@pyai/sdk 0.16.0 → 0.17.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +2 -2
- package/dist/cli-omni-template.d.ts +2 -2
- package/dist/cli-omni-template.js +2 -2
- package/dist/dialer/dialerElement.js +1 -1
- package/dist/dialer/managedDialer.d.ts +2 -1
- package/dist/dialer/managedDialer.js +4 -2
- package/dist/dialer/serverBroker.js +10 -5
- package/dist/index.d.ts +102 -3
- package/dist/index.js +12 -1
- package/package.json +1 -1
- package/src/cli-omni-template.ts +2 -2
- package/src/dialer/dialerElement.ts +1 -1
- package/src/dialer/managedDialer.ts +5 -3
- package/src/dialer/serverBroker.ts +9 -4
- package/src/index.ts +77 -3
package/README.md
CHANGED
|
@@ -106,12 +106,12 @@ the instant the socket opens and routes server frames to typed callbacks, so you
|
|
|
106
106
|
is acked but silently dropped, giving you a connected session with zero turns):
|
|
107
107
|
|
|
108
108
|
<!-- omni-starter:start -->
|
|
109
|
-
Requires SDK 0.
|
|
109
|
+
Requires SDK 0.17.0 for `--template omni`. The runnable starter uses ESM,
|
|
110
110
|
Node 20.19+ with an explicit `ws` transport, or Node 22+. MCP 0.5.1
|
|
111
111
|
requires Node 22+. The SDK's REST client continues to support Node 18+.
|
|
112
112
|
|
|
113
113
|
```sh
|
|
114
|
-
npm install @pyai/sdk@0.
|
|
114
|
+
npm install @pyai/sdk@0.17.0
|
|
115
115
|
npx pyai init voice-demo --template omni
|
|
116
116
|
cd voice-demo
|
|
117
117
|
npm install
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
export declare const omniTemplate: Readonly<{
|
|
2
|
-
"README.md": "# Runnable Omni starter\n\nRequires Node 20.19+ and ESM. Node 20 uses the included `ws` transport;\nNode 22 also works. The separate MCP server requires Node 22+.\nThis server-side example uses a synthetic read-only office-hours tool. Read\n`PYAI.md` when generating the project through the CLI.\n\n<!-- omni-install:start -->\n```sh\nnpm install @pyai/sdk@0.
|
|
3
|
-
"package.json": "{\n \"name\": \"pyai-omni-starter\",\n \"version\": \"0.0.0\",\n \"private\": true,\n \"type\": \"module\",\n \"engines\": {\n \"node\": \">=20.19.0\"\n },\n \"scripts\": {\n \"start\": \"node main.mjs\"\n },\n \"dependencies\": {\n \"@pyai/sdk\": \"0.
|
|
2
|
+
"README.md": "# Runnable Omni starter\n\nRequires Node 20.19+ and ESM. Node 20 uses the included `ws` transport;\nNode 22 also works. The separate MCP server requires Node 22+.\nThis server-side example uses a synthetic read-only office-hours tool. Read\n`PYAI.md` when generating the project through the CLI.\n\n<!-- omni-install:start -->\n```sh\nnpm install @pyai/sdk@0.17.0\nnpx pyai init voice-demo --template omni\ncd voice-demo\nnpm install\n```\n\nInitialization is offline; installing dependencies and running `main.mjs` are\nseparate steps. For an existing checkout of this example, run `npm install`\nin its directory instead.\n<!-- omni-install:end -->\n\nInject `PYAI_API_KEY` through your secret manager or shell environment. Do not\npaste a real key into source, a prompt, or shell history. `.env.example` documents\nthe variable; this program does not load `.env` or saved CLI profiles.\n\nPrepare a 24 kHz PCM16 mono WAV, at most 20 seconds, saying “Please look up the\noffice opening time.” Then run:\n\n```sh\nnode main.mjs caller.wav\n```\n\nRunning the program explicitly opens one bounded Omni session and uploads its\ncaptured reply to Hear. Both consume your key's allowed usage. Automatic retries\nare disabled. No managed number is bought and no phone call is placed.\n\n`session.mjs` waits for configuration and the greeting to drain, sends exactly\none paced input stream (caller PCM or silence), returns the read tool's result,\nand captures output using the rate from `hello.audio_out`. Never run a separate\nsilence timer alongside caller audio. `timing.mjs` is generated from the existing\nOmni evaluation harness's timing utilities.\n\nThe new private `omni-run-*` folder contains `session.wav`, `reply.wav`, and\n`report.json`. The report separates received audio, an office-hours answer\nrecovered by Hear from captured audio, and physical playback (not tested).\nThe office-hours check accepts formatting such as `opens@9a.m`, `9 a.m.` and\n`nine in the morning` in a complete affirmative opening-time sentence. Wrong\ntimes, negation, conflicting or truncated text remain unverified. This is a\nconservative check for the example fact, not a general answer-quality score.\nA synthesis transcript is advisory; it is not evidence that speech arrived.\nThis starter does not negotiate playback receipts. Capture ends after the\nsimulated queue drains and two seconds of quiet, a bounded capture heuristic.\nListen to the saved file when qualifying actual sound quality or completeness.\n\nTo exercise interruption, supply a second WAV saying “Stop speaking now”:\n\n```sh\nnode main.mjs caller.wav interruption.wav\n```\n\nThe second clip starts while reply audio is queued. `onBargeIn` clears that\nsimulated queue, and capture observes at least five seconds after that clip ends.\nAn interrupted answer can correctly fail the full-answer check;\ninspect the separate interruption evidence. This is not a physical speaker test.\nFor a real playback adapter, cancel both queued and currently playing audio.\nReplace the example lookup with authorized application data before product use.\n\nOn a session error, the program exits with JSON containing a recognized machine\n`code` (for example `media_dead` or `invalid_configure`). Unknown server codes\nbecome `unrecognized_server_error`; transport or decoding errors without a\nrecognized transport code become `transport_or_protocol_error`. Raw error text,\ncall IDs and unknown code values are omitted. No automatic retry is attempted;\ninspect the code and your authorized diagnostics before deciding to retry.\n";
|
|
3
|
+
"package.json": "{\n \"name\": \"pyai-omni-starter\",\n \"version\": \"0.0.0\",\n \"private\": true,\n \"type\": \"module\",\n \"engines\": {\n \"node\": \">=20.19.0\"\n },\n \"scripts\": {\n \"start\": \"node main.mjs\"\n },\n \"dependencies\": {\n \"@pyai/sdk\": \"0.17.0\",\n \"ws\": \"^8.18.3\"\n }\n}\n";
|
|
4
4
|
"timing.mjs": "// Monotonic capture timing. Arrival time is distinct from queued playback:\n// a single packet may contain seconds of audio and must finish before settling.\nexport const FRAME_MS = 20;\nexport const REALTIME_GAP_LIMIT_MS = 100;\n// Interruption tests need an observation window after caller transmission,\n// even when an earlier cue/response already finished. These are test bounds,\n// not an inferred engine reply-end or a naturalness threshold.\nexport const INTERRUPTION_OBSERVATION = Object.freeze({\n minimumAfterCallerStreamMs: 5000, quietMs: 2000, maximumAfterCallerStreamMs: 25000,\n});\n\n/** Approximate audible bounds; this is energy detection, not a speech/VAD claim. */\nexport function audibleBounds(pcm, rate, rmsFloor = 160) {\n const window = Math.max(1, Math.round(rate / 100)); // 10 ms windows\n let first = null;\n let last = null;\n for (let off = 0; off < pcm.length; off += window) {\n const end = Math.min(pcm.length, off + window);\n let energy = 0;\n for (let i = off; i < end; i++) energy += pcm[i] * pcm[i];\n if (Math.sqrt(energy / (end - off)) >= rmsFloor) {\n first ??= off;\n last = end;\n }\n }\n return { first, last };\n}\n\nexport function newAudioCapture(started) {\n return {\n started, firstPacketAt: null, lastPacketAt: null,\n firstAudioAt: null, lastAudioAt: null, playbackEndAt: null,\n samples: 0, pcm: [], turnBeginAt: null, eouMs: null,\n latestTurnBeginAt: null, postTurnBeginFirstPacketAt: null,\n postTurnBeginFirstAudioAt: null, lastAssistantTranscriptAt: null, lastCallerTranscriptAt: null,\n callerTranscriptEvents: [], assistantTranscriptEvents: [],\n turnBegins: [], unheardCancellations: [], unheardSupersededTurns: [], tools: [], toolResults: [], kb: null, events: [],\n };\n}\n\nexport function recordTurnBegin(ctx, at, turn, origin = 0) {\n const prior = ctx.turnBegins.at(-1);\n const cancellation = ctx.unheardCancellations.at(-1);\n if (prior && cancellation && !cancellation.admitted\n && cancellation.turn === prior.turn && cancellation.turnBeginIndex === ctx.turnBegins.length - 1\n && Number.isSafeInteger(turn) && turn > prior.turn && at > cancellation.at\n && ctx.samples === 0 && ctx.assistantTranscriptEvents.length === 0\n && ctx.tools.length === 0 && ctx.toolResults.length === 0) {\n cancellation.admitted = true;\n ctx.unheardSupersededTurns.push({ turn: prior.turn, beginAtMs: prior.atMs,\n cancelledAtMs: cancellation.atMs, cancellationEventIndex: cancellation.eventIndex,\n replacementTurn: turn, outputPacketsBeforeReplacement: 0 });\n }\n ctx.turnBeginAt ??= at;\n ctx.latestTurnBeginAt = at;\n // Earlier PCM can be a listening cue. A new reply is still pending even\n // when that cue has already finished playing; retain all of its PCM.\n ctx.postTurnBeginFirstPacketAt = null;\n ctx.postTurnBeginFirstAudioAt = null;\n ctx.turnBegins.push({ atMs: Math.round(at - origin), turn: turn ?? null });\n}\n\nexport function recordUnheardCancellation(ctx, event, at, origin, eventIndex) {\n const prior = ctx.turnBegins.at(-1);\n if (event.event !== \"flush\" || event.reason !== \"turn_merge\"\n || !Number.isSafeInteger(event.cancelled_turn) || event.cancelled_turn <= 0\n || prior?.turn !== event.cancelled_turn || at < ctx.latestTurnBeginAt\n || ctx.samples !== 0 || ctx.assistantTranscriptEvents.length\n || ctx.tools.length || ctx.toolResults.length\n || ctx.unheardCancellations.some(c => c.turnBeginIndex === ctx.turnBegins.length - 1)) return;\n ctx.unheardCancellations.push({ turn: event.cancelled_turn, at,\n atMs: Math.round(at - origin), eventIndex, turnBeginIndex: ctx.turnBegins.length - 1,\n admitted: false });\n}\n\nexport function recordAudio(ctx, pcm, at, rate) {\n if (!(pcm instanceof Int16Array) || !pcm.length) return null;\n const copy = pcm.slice(); // Some transports reuse callback buffers.\n const playbackAt = Math.max(at, ctx.playbackEndAt ?? at);\n const endAt = playbackAt + (copy.length / rate) * 1000;\n const bounds = audibleBounds(copy, rate);\n ctx.firstPacketAt ??= at;\n ctx.lastPacketAt = at;\n ctx.playbackEndAt = endAt;\n const afterTurnBegin = ctx.latestTurnBeginAt != null && at >= ctx.latestTurnBeginAt;\n if (afterTurnBegin) ctx.postTurnBeginFirstPacketAt ??= at;\n if (bounds.first != null) {\n ctx.firstAudioAt ??= playbackAt + (bounds.first / rate) * 1000;\n ctx.lastAudioAt = playbackAt + (bounds.last / rate) * 1000;\n if (afterTurnBegin) ctx.postTurnBeginFirstAudioAt ??= playbackAt + (bounds.first / rate) * 1000;\n }\n ctx.samples += copy.length;\n ctx.pcm.push(copy);\n return { playbackAt, endAt, pcm: copy };\n}\n\n/** Preserve negative gaps: they are overlaps, never missing measurements. */\nexport function captureTiming(ctx, caller, origin = 0) {\n const relative = (at) => at == null ? null : Math.round(at - origin);\n const gap = (at) => at == null || caller.speechOffsetAt == null\n ? null : Math.round(at - caller.speechOffsetAt);\n return {\n ttfbMs: gap(ctx.firstAudioAt),\n anyAudioTtfbMs: gap(ctx.firstAudioAt),\n postTurnBeginTtfbMs: gap(ctx.postTurnBeginFirstAudioAt),\n turnMs: gap(ctx.lastAudioAt),\n packetTtfbMs: gap(ctx.firstPacketAt),\n callerStartMs: relative(caller.startedAt),\n callerSpeechOnsetMs: relative(caller.speechOnsetAt),\n callerSpeechOffsetMs: relative(caller.speechOffsetAt),\n callerStreamEndMs: relative(caller.streamEndAt),\n agentFirstPacketMs: relative(ctx.firstPacketAt),\n agentLastPacketMs: relative(ctx.lastPacketAt),\n agentSpeechOnsetMs: relative(ctx.firstAudioAt),\n agentSpeechOffsetMs: relative(ctx.lastAudioAt),\n agentPlaybackEndMs: relative(ctx.playbackEndAt),\n latestTurnBeginMs: relative(ctx.latestTurnBeginAt),\n postTurnBeginFirstPacketMs: relative(ctx.postTurnBeginFirstPacketAt),\n postTurnBeginSpeechOnsetMs: relative(ctx.postTurnBeginFirstAudioAt),\n lastAssistantTranscriptMs: relative(ctx.lastAssistantTranscriptAt),\n callerMaxFrameGapMs: Math.round(caller.maxFrameGapMs ?? 0),\n method: \"client-monotonic-queued-playback-energy-bounds\",\n };\n}\n\nexport async function streamPcmRealtime(omni, pcm, rate, clock, onFrame = () => {}) {\n const frameSize = Math.max(1, Math.round((rate * FRAME_MS) / 1000));\n const bounds = audibleBounds(pcm, rate);\n const startedAt = clock.now();\n let speechOnsetAt = null;\n let speechOffsetAt = null;\n let previousFrameEnd = startedAt;\n let maxFrameGapMs = 0;\n for (let off = 0; off < pcm.length; off += frameSize) {\n const frame = pcm.subarray(off, Math.min(off + frameSize, pcm.length));\n const at = clock.now();\n maxFrameGapMs = Math.max(maxFrameGapMs, at - previousFrameEnd);\n const durationMs = (frame.length / rate) * 1000;\n omni.sendAudio(frame);\n onFrame(frame, at);\n if (bounds.first != null && bounds.first >= off && bounds.first < off + frame.length) {\n speechOnsetAt = at + ((bounds.first - off) / rate) * 1000;\n }\n if (bounds.last != null && bounds.last > off && bounds.last <= off + frame.length) {\n speechOffsetAt = at + ((bounds.last - off) / rate) * 1000;\n }\n previousFrameEnd = at + durationMs;\n // Pace from the actual send. Never burst late frames to \"catch up\".\n await clock.sleep(Math.max(0, previousFrameEnd - clock.now()));\n }\n return { startedAt, speechOnsetAt, speechOffsetAt,\n streamEndAt: clock.now(), maxFrameGapMs };\n}\n\nexport async function streamSilenceWhile(omni, rate, shouldContinue, clock, onFrame = () => {}) {\n const silence = new Int16Array(Math.max(1, Math.round((rate * FRAME_MS) / 1000)));\n while (shouldContinue()) {\n const at = clock.now();\n omni.sendAudio(silence);\n onFrame(silence, at);\n await clock.sleep(FRAME_MS);\n }\n}\n\nexport async function waitForAgentSettle(getCtx, clock, opts = {}) {\n const started = clock.now();\n const observation = opts.callerStreamEndAt != null;\n if (observation && (!Number.isFinite(opts.callerStreamEndAt)\n || opts.callerStreamEndAt < 0 || opts.callerStreamEndAt > started)) {\n throw new Error(\"Invalid caller observation boundary\");\n }\n const timeoutMs = observation ? INTERRUPTION_OBSERVATION.maximumAfterCallerStreamMs : opts.timeoutMs ?? 25000;\n const settleMs = observation ? INTERRUPTION_OBSERVATION.quietMs : opts.settleMs ?? 2000;\n const deadlineAt = (observation ? opts.callerStreamEndAt : started) + timeoutMs;\n const minimumAt = observation ? opts.callerStreamEndAt + INTERRUPTION_OBSERVATION.minimumAfterCallerStreamMs : -Infinity;\n while (true) {\n const ctx = getCtx();\n const at = clock.now();\n if (opts.isClosed?.()) return { reason: \"closed\", at };\n if (observation && at >= deadlineAt) return { reason: \"timeout\", at };\n const hasResponseAudio = !opts.requireTurnBegin || (ctx.latestTurnBeginAt != null\n && ctx.postTurnBeginFirstAudioAt != null);\n // Synthesis text is progress, not a completion event. It can arrive after\n // early/final audio, so extend the quiet floor without demanding another\n // packet after every advisory. The protocol still has no reply-end marker.\n const activityEnd = Math.max(ctx.playbackEndAt ?? -Infinity,\n ctx.latestTurnBeginAt ?? -Infinity, ctx.lastAssistantTranscriptAt ?? -Infinity,\n observation ? ctx.lastCallerTranscriptAt ?? -Infinity : -Infinity);\n // Do not demand a new post-caller turn_begin: an earlier turn can be\n // coalesced and speak later. This records a bounded quiet observation;\n // downstream response-timing checks can impose a stricter begin boundary.\n if (ctx.firstPacketAt != null && hasResponseAudio && at >= minimumAt && at >= activityEnd + settleMs) {\n return { reason: \"settled\", at };\n }\n if (opts.allowEmpty && ctx.firstPacketAt == null && at >= minimumAt && at - started >= (opts.emptyWaitMs ?? 800)) {\n return { reason: \"empty\", at };\n }\n if (at - started >= timeoutMs) return { reason: \"timeout\", at };\n await clock.sleep(observation ? Math.min(20, deadlineAt - at) : 20);\n }\n}\n\nexport async function withTimeout(promise, ms, message, clock) {\n let timer;\n try {\n return await Promise.race([\n promise,\n new Promise((_, reject) => {\n timer = clock.setTimeout(() => reject(new Error(message)), ms);\n }),\n ]);\n } finally {\n clock.clearTimeout(timer);\n }\n}\n";
|
|
5
5
|
"main.mjs": "import PyAI from \"@pyai/sdk\";\nimport WebSocket from \"ws\"; // Explicit transport makes this work on Node 20 too.\nimport { readFile, writeFile, mkdtemp } from \"node:fs/promises\";\nimport { join } from \"node:path\";\nimport { runSession } from \"./session.mjs\";\nimport { readWav, writeWav, assessAnswer } from \"./audio.mjs\";\n\nasync function main() {\n const [callerPath, interruptionPath, extra] = process.argv.slice(2);\n if (!callerPath || extra) throw new Error(\"Usage: node main.mjs caller.wav [interruption.wav]\");\n if (!process.env.PYAI_API_KEY) throw new Error(\"Inject PYAI_API_KEY through your environment first\");\n const caller = readWav(await readFile(callerPath));\n const interruption = interruptionPath ? readWav(await readFile(interruptionPath)) : undefined;\n const pyai = new PyAI({ apiKey: process.env.PYAI_API_KEY, maxRetries: 0,\n fetch: (url, options) => fetch(url, { ...options, signal: AbortSignal.timeout(30000) }),\n });\n const { report, audio, replyAudio } = await runSession({ pyai, webSocket: WebSocket, caller, interruption });\n const directory = await mkdtemp(join(process.cwd(), \"omni-run-\"));\n const save = (name, data) => writeFile(join(directory, name), data, { mode: 0o600, flag: \"wx\" });\n await save(\"session.wav\", writeWav(audio, report.output_rate ?? 24000));\n const reply = writeWav(replyAudio, report.output_rate ?? 24000);\n await save(\"reply.wav\", reply);\n // Hear the captured bytes, rather than trusting synthesis-advisory text.\n let capturedTranscript = \"\";\n if (report.reply_audio_bytes) {\n try {\n const heard = await pyai.audio.transcriptions.create({ file: new File([reply], \"reply.wav\", { type: \"audio/wav\" }), model: \"pyai-hear\" });\n capturedTranscript = heard.text;\n } catch { report.hear_error = \"Captured audio saved; Hear verification failed\"; }\n }\n const result = { ...report, captured_transcript: capturedTranscript, evidence: assessAnswer(report, capturedTranscript) };\n await save(\"report.json\", JSON.stringify(result, null, 2) + \"\\n\");\n console.log(JSON.stringify({ directory, ...result }, null, 2));\n if (result.evidence.captured_answer !== \"verified_by_hear\" || (interruption && result.evidence.interruption !== \"simulated_queue_cleared\")) process.exitCode = 1;\n}\n\nmain().catch(error => {\n const message = String(error.message).replaceAll(process.env.PYAI_API_KEY || \"\\0\", \"[REDACTED]\");\n console.error(JSON.stringify({ error: message, ...(error.omni_code ? { code: error.omni_code } : {}) })); process.exitCode = 1;\n});\n";
|
|
6
6
|
"session.mjs": "import { newAudioCapture, recordAudio, recordTurnBegin, streamPcmRealtime, INTERRUPTION_OBSERVATION } from \"./timing.mjs\";\n\nexport const realClock = {\n now: () => performance.now(),\n sleep: ms => new Promise(resolve => setTimeout(resolve, ms)),\n};\n\n// Only recognized machine codes may leave the example. Error messages, call\n// IDs and arbitrary code values can contain credentials or customer content.\nconst publicErrorCodes = new Set([\n \"invalid_configure\", \"unsupported_tool_transport\", \"tool_configuration_locked\",\n \"media_dead\", \"unauthorized\", \"insufficient_scope\", \"rate_limit_exceeded\",\n \"concurrency_limit_exceeded\", \"daily_cap_exceeded\", \"credit_exhausted\",\n]);\nconst transportErrorCodes = new Set([\"ECONNRESET\", \"ECONNREFUSED\", \"ETIMEDOUT\", \"ENOTFOUND\", \"EAI_AGAIN\"]);\nexport function omniFailure(error) {\n const code = error instanceof Error\n ? (transportErrorCodes.has(error.code) ? error.code : \"transport_or_protocol_error\")\n : (publicErrorCodes.has(error?.code) ? error.code : \"unrecognized_server_error\");\n return Object.assign(new Error(`Omni failed (${code}); no automatic retry was attempted`), { omni_code: code });\n}\n\n/** One bounded, server-side example. Its playback queue is simulated. */\nexport async function runSession({ pyai, webSocket, caller, interruption, clock = realClock, timeoutMs = 60000 }) {\n const rate = 24000, frameSize = 480, silence = new Int16Array(frameSize);\n if (!(caller instanceof Int16Array) || !caller.length || caller.length > rate * 20) throw new Error(\"Provide 1–20 seconds of caller PCM16 at 24 kHz\");\n if (interruption && (!(interruption instanceof Int16Array) || !interruption.length || interruption.length > rate * 20)) throw new Error(\"Invalid interruption PCM\");\n const started = clock.now(), all = newAudioCapture(started), reply = newAudioCapture(started);\n const report = { input_rate: rate, output_rate: null, playback_sink: \"simulated\", tool_calls: 0, tool_executions: 0, duplicate_tool_calls: 0, flushes_after_interruption: 0, cleared_queue_ms: 0, interruption_requested: !!interruption, caller_started_ms: null, interruption_started_ms: null, advisory_transcripts: [], end_reason: \"timeout\" };\n let session, configuredAt = null, closed = false, failure = null, phase = \"greeting\", offset = 0, queueEnd = started, interruptionSent = false, inputEndAt = started;\n const handled = new Set();\n const fail = message => { failure ??= new Error(message); };\n try {\n session = pyai.omni.connect({\n webSocket, rate,\n configure: {\n voice_id: \"stock_amos_en_us\", language: \"en\", greeting: \"Hello. How can I help?\",\n persona: \"When asked about office hours, call lookup_office_hours. Never guess. After the result, say one short sentence giving its opening_time.\",\n tools: [{ name: \"lookup_office_hours\", description: \"Read the synthetic example office opening time.\", side_effect: \"read\", parameters: { type: \"object\", properties: {}, required: [], additionalProperties: false } }],\n },\n onHello: frame => {\n const match = /^pcm16@(8000|24000)$/.exec(frame.audio_out ?? \"\");\n if (!match) return fail(\"Unsupported Omni output format\");\n report.output_rate = Number(match[1]);\n },\n onConfigured: () => { configuredAt ??= clock.now(); },\n onAudio: chunk => {\n if (!report.output_rate) return fail(\"Audio arrived before its output format\");\n const bytes = chunk instanceof ArrayBuffer ? new Uint8Array(chunk) : new Uint8Array(chunk.buffer, chunk.byteOffset, chunk.byteLength);\n if (bytes.byteLength % 2) return fail(\"Invalid PCM16 output\");\n const view = new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength);\n const pcm = Int16Array.from({ length: bytes.length / 2 }, (_, i) => view.getInt16(i * 2, true));\n const now = clock.now();\n recordAudio(all, pcm, now, report.output_rate);\n queueEnd = Math.max(now, queueEnd) + pcm.length / report.output_rate * 1000;\n if (report.tool_executions) recordAudio(reply, pcm, now, report.output_rate);\n },\n onEvent: frame => {\n if (frame.event === \"turn_begin\") recordTurnBegin(all, clock.now(), frame.turn);\n },\n onTranscript: frame => {\n // An assistant advisory is progress; never treat it as received speech.\n if (report.advisory_transcripts.length < 100) report.advisory_transcripts.push({ role: frame.role, text: frame.text });\n if (frame.role === \"assistant\") all.lastAssistantTranscriptAt = clock.now();\n },\n onToolCall: frame => {\n report.tool_calls++;\n if (handled.has(frame.call_id)) { report.duplicate_tool_calls++; return; }\n handled.add(frame.call_id);\n if (frame.name !== \"lookup_office_hours\") { session.toolResult(frame.call_id, { error: \"Unknown example tool\" }); return; }\n // Replace this read-only fixture with your authorized data lookup.\n report.tool_executions++;\n session.toolResult(frame.call_id, { result: { opening_time: \"9 a.m.\", source: \"synthetic example office\" } });\n },\n onBargeIn: () => {\n const now = clock.now(), queued = Math.max(0, queueEnd - now);\n if (report.interruption_started_ms != null) { report.flushes_after_interruption++; report.cleared_queue_ms += queued; }\n // A real speaker adapter must cancel scheduled and currently playing audio.\n queueEnd = now; all.playbackEndAt = now; reply.playbackEndAt = now;\n },\n onError: error => { failure ??= omniFailure(error); },\n onClose: () => { closed = true; },\n });\n while (clock.now() - started < timeoutMs) {\n if (failure) throw failure;\n if (closed) { report.end_reason = \"closed\"; break; }\n const now = clock.now();\n if (configuredAt == null || !report.output_rate) {\n if (now - started >= 10000) throw new Error(\"Omni connection/configuration timed out\");\n await clock.sleep(20); continue;\n }\n const quietAt = Math.max(queueEnd, all.lastPacketAt ?? now, all.lastAssistantTranscriptAt ?? 0, inputEndAt);\n if (phase === \"greeting\") {\n if (all.firstPacketAt != null && now - configuredAt >= 2000 && now >= quietAt + 800) {\n phase = \"caller\"; report.caller_started_ms = now - started;\n } else if (now - configuredAt > 15000) throw new Error(\"Greeting did not drain within 15 seconds\");\n }\n if (phase === \"reply\" && interruption && !interruptionSent && reply.samples && queueEnd - now >= 200) {\n phase = \"interruption\"; offset = 0; interruptionSent = true; report.interruption_started_ms = now - started;\n }\n const observationEnd = inputEndAt + (interruptionSent ? INTERRUPTION_OBSERVATION.minimumAfterCallerStreamMs : 0);\n if (phase === \"reply\" && reply.samples && now >= quietAt + 2000 && now >= observationEnd && (!interruption || interruptionSent)) {\n report.end_reason = \"quiet_window\"; break;\n }\n // Exactly one producer: each 20 ms slot contains caller PCM OR silence.\n let pcm = silence;\n if (phase === \"caller\" || phase === \"interruption\") {\n const source = phase === \"caller\" ? caller : interruption;\n pcm = source.subarray(offset, offset + frameSize); offset += pcm.length;\n if (offset >= source.length) {\n inputEndAt = now + pcm.length / rate * 1000;\n report[phase === \"caller\" ? \"caller_completed_ms\" : \"interruption_completed_ms\"] = inputEndAt - started;\n phase = \"reply\"; offset = 0;\n }\n }\n await streamPcmRealtime(session, pcm, rate, clock);\n }\n report.audio_bytes = all.samples * 2;\n report.reply_audio_bytes = reply.samples * 2;\n report.elapsed_ms = Math.round(clock.now() - started);\n return { report, audio: all.pcm, replyAudio: reply.pcm };\n } finally { session?.close(); }\n}\n";
|
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
// Generated by scripts/generate-omni-starter.mjs; edit examples/omni-starter instead.
|
|
2
2
|
export const omniTemplate = Object.freeze({
|
|
3
|
-
"README.md": "# Runnable Omni starter\n\nRequires Node 20.19+ and ESM. Node 20 uses the included `ws` transport;\nNode 22 also works. The separate MCP server requires Node 22+.\nThis server-side example uses a synthetic read-only office-hours tool. Read\n`PYAI.md` when generating the project through the CLI.\n\n<!-- omni-install:start -->\n```sh\nnpm install @pyai/sdk@0.
|
|
4
|
-
"package.json": "{\n \"name\": \"pyai-omni-starter\",\n \"version\": \"0.0.0\",\n \"private\": true,\n \"type\": \"module\",\n \"engines\": {\n \"node\": \">=20.19.0\"\n },\n \"scripts\": {\n \"start\": \"node main.mjs\"\n },\n \"dependencies\": {\n \"@pyai/sdk\": \"0.
|
|
3
|
+
"README.md": "# Runnable Omni starter\n\nRequires Node 20.19+ and ESM. Node 20 uses the included `ws` transport;\nNode 22 also works. The separate MCP server requires Node 22+.\nThis server-side example uses a synthetic read-only office-hours tool. Read\n`PYAI.md` when generating the project through the CLI.\n\n<!-- omni-install:start -->\n```sh\nnpm install @pyai/sdk@0.17.0\nnpx pyai init voice-demo --template omni\ncd voice-demo\nnpm install\n```\n\nInitialization is offline; installing dependencies and running `main.mjs` are\nseparate steps. For an existing checkout of this example, run `npm install`\nin its directory instead.\n<!-- omni-install:end -->\n\nInject `PYAI_API_KEY` through your secret manager or shell environment. Do not\npaste a real key into source, a prompt, or shell history. `.env.example` documents\nthe variable; this program does not load `.env` or saved CLI profiles.\n\nPrepare a 24 kHz PCM16 mono WAV, at most 20 seconds, saying “Please look up the\noffice opening time.” Then run:\n\n```sh\nnode main.mjs caller.wav\n```\n\nRunning the program explicitly opens one bounded Omni session and uploads its\ncaptured reply to Hear. Both consume your key's allowed usage. Automatic retries\nare disabled. No managed number is bought and no phone call is placed.\n\n`session.mjs` waits for configuration and the greeting to drain, sends exactly\none paced input stream (caller PCM or silence), returns the read tool's result,\nand captures output using the rate from `hello.audio_out`. Never run a separate\nsilence timer alongside caller audio. `timing.mjs` is generated from the existing\nOmni evaluation harness's timing utilities.\n\nThe new private `omni-run-*` folder contains `session.wav`, `reply.wav`, and\n`report.json`. The report separates received audio, an office-hours answer\nrecovered by Hear from captured audio, and physical playback (not tested).\nThe office-hours check accepts formatting such as `opens@9a.m`, `9 a.m.` and\n`nine in the morning` in a complete affirmative opening-time sentence. Wrong\ntimes, negation, conflicting or truncated text remain unverified. This is a\nconservative check for the example fact, not a general answer-quality score.\nA synthesis transcript is advisory; it is not evidence that speech arrived.\nThis starter does not negotiate playback receipts. Capture ends after the\nsimulated queue drains and two seconds of quiet, a bounded capture heuristic.\nListen to the saved file when qualifying actual sound quality or completeness.\n\nTo exercise interruption, supply a second WAV saying “Stop speaking now”:\n\n```sh\nnode main.mjs caller.wav interruption.wav\n```\n\nThe second clip starts while reply audio is queued. `onBargeIn` clears that\nsimulated queue, and capture observes at least five seconds after that clip ends.\nAn interrupted answer can correctly fail the full-answer check;\ninspect the separate interruption evidence. This is not a physical speaker test.\nFor a real playback adapter, cancel both queued and currently playing audio.\nReplace the example lookup with authorized application data before product use.\n\nOn a session error, the program exits with JSON containing a recognized machine\n`code` (for example `media_dead` or `invalid_configure`). Unknown server codes\nbecome `unrecognized_server_error`; transport or decoding errors without a\nrecognized transport code become `transport_or_protocol_error`. Raw error text,\ncall IDs and unknown code values are omitted. No automatic retry is attempted;\ninspect the code and your authorized diagnostics before deciding to retry.\n",
|
|
4
|
+
"package.json": "{\n \"name\": \"pyai-omni-starter\",\n \"version\": \"0.0.0\",\n \"private\": true,\n \"type\": \"module\",\n \"engines\": {\n \"node\": \">=20.19.0\"\n },\n \"scripts\": {\n \"start\": \"node main.mjs\"\n },\n \"dependencies\": {\n \"@pyai/sdk\": \"0.17.0\",\n \"ws\": \"^8.18.3\"\n }\n}\n",
|
|
5
5
|
"timing.mjs": "// Monotonic capture timing. Arrival time is distinct from queued playback:\n// a single packet may contain seconds of audio and must finish before settling.\nexport const FRAME_MS = 20;\nexport const REALTIME_GAP_LIMIT_MS = 100;\n// Interruption tests need an observation window after caller transmission,\n// even when an earlier cue/response already finished. These are test bounds,\n// not an inferred engine reply-end or a naturalness threshold.\nexport const INTERRUPTION_OBSERVATION = Object.freeze({\n minimumAfterCallerStreamMs: 5000, quietMs: 2000, maximumAfterCallerStreamMs: 25000,\n});\n\n/** Approximate audible bounds; this is energy detection, not a speech/VAD claim. */\nexport function audibleBounds(pcm, rate, rmsFloor = 160) {\n const window = Math.max(1, Math.round(rate / 100)); // 10 ms windows\n let first = null;\n let last = null;\n for (let off = 0; off < pcm.length; off += window) {\n const end = Math.min(pcm.length, off + window);\n let energy = 0;\n for (let i = off; i < end; i++) energy += pcm[i] * pcm[i];\n if (Math.sqrt(energy / (end - off)) >= rmsFloor) {\n first ??= off;\n last = end;\n }\n }\n return { first, last };\n}\n\nexport function newAudioCapture(started) {\n return {\n started, firstPacketAt: null, lastPacketAt: null,\n firstAudioAt: null, lastAudioAt: null, playbackEndAt: null,\n samples: 0, pcm: [], turnBeginAt: null, eouMs: null,\n latestTurnBeginAt: null, postTurnBeginFirstPacketAt: null,\n postTurnBeginFirstAudioAt: null, lastAssistantTranscriptAt: null, lastCallerTranscriptAt: null,\n callerTranscriptEvents: [], assistantTranscriptEvents: [],\n turnBegins: [], unheardCancellations: [], unheardSupersededTurns: [], tools: [], toolResults: [], kb: null, events: [],\n };\n}\n\nexport function recordTurnBegin(ctx, at, turn, origin = 0) {\n const prior = ctx.turnBegins.at(-1);\n const cancellation = ctx.unheardCancellations.at(-1);\n if (prior && cancellation && !cancellation.admitted\n && cancellation.turn === prior.turn && cancellation.turnBeginIndex === ctx.turnBegins.length - 1\n && Number.isSafeInteger(turn) && turn > prior.turn && at > cancellation.at\n && ctx.samples === 0 && ctx.assistantTranscriptEvents.length === 0\n && ctx.tools.length === 0 && ctx.toolResults.length === 0) {\n cancellation.admitted = true;\n ctx.unheardSupersededTurns.push({ turn: prior.turn, beginAtMs: prior.atMs,\n cancelledAtMs: cancellation.atMs, cancellationEventIndex: cancellation.eventIndex,\n replacementTurn: turn, outputPacketsBeforeReplacement: 0 });\n }\n ctx.turnBeginAt ??= at;\n ctx.latestTurnBeginAt = at;\n // Earlier PCM can be a listening cue. A new reply is still pending even\n // when that cue has already finished playing; retain all of its PCM.\n ctx.postTurnBeginFirstPacketAt = null;\n ctx.postTurnBeginFirstAudioAt = null;\n ctx.turnBegins.push({ atMs: Math.round(at - origin), turn: turn ?? null });\n}\n\nexport function recordUnheardCancellation(ctx, event, at, origin, eventIndex) {\n const prior = ctx.turnBegins.at(-1);\n if (event.event !== \"flush\" || event.reason !== \"turn_merge\"\n || !Number.isSafeInteger(event.cancelled_turn) || event.cancelled_turn <= 0\n || prior?.turn !== event.cancelled_turn || at < ctx.latestTurnBeginAt\n || ctx.samples !== 0 || ctx.assistantTranscriptEvents.length\n || ctx.tools.length || ctx.toolResults.length\n || ctx.unheardCancellations.some(c => c.turnBeginIndex === ctx.turnBegins.length - 1)) return;\n ctx.unheardCancellations.push({ turn: event.cancelled_turn, at,\n atMs: Math.round(at - origin), eventIndex, turnBeginIndex: ctx.turnBegins.length - 1,\n admitted: false });\n}\n\nexport function recordAudio(ctx, pcm, at, rate) {\n if (!(pcm instanceof Int16Array) || !pcm.length) return null;\n const copy = pcm.slice(); // Some transports reuse callback buffers.\n const playbackAt = Math.max(at, ctx.playbackEndAt ?? at);\n const endAt = playbackAt + (copy.length / rate) * 1000;\n const bounds = audibleBounds(copy, rate);\n ctx.firstPacketAt ??= at;\n ctx.lastPacketAt = at;\n ctx.playbackEndAt = endAt;\n const afterTurnBegin = ctx.latestTurnBeginAt != null && at >= ctx.latestTurnBeginAt;\n if (afterTurnBegin) ctx.postTurnBeginFirstPacketAt ??= at;\n if (bounds.first != null) {\n ctx.firstAudioAt ??= playbackAt + (bounds.first / rate) * 1000;\n ctx.lastAudioAt = playbackAt + (bounds.last / rate) * 1000;\n if (afterTurnBegin) ctx.postTurnBeginFirstAudioAt ??= playbackAt + (bounds.first / rate) * 1000;\n }\n ctx.samples += copy.length;\n ctx.pcm.push(copy);\n return { playbackAt, endAt, pcm: copy };\n}\n\n/** Preserve negative gaps: they are overlaps, never missing measurements. */\nexport function captureTiming(ctx, caller, origin = 0) {\n const relative = (at) => at == null ? null : Math.round(at - origin);\n const gap = (at) => at == null || caller.speechOffsetAt == null\n ? null : Math.round(at - caller.speechOffsetAt);\n return {\n ttfbMs: gap(ctx.firstAudioAt),\n anyAudioTtfbMs: gap(ctx.firstAudioAt),\n postTurnBeginTtfbMs: gap(ctx.postTurnBeginFirstAudioAt),\n turnMs: gap(ctx.lastAudioAt),\n packetTtfbMs: gap(ctx.firstPacketAt),\n callerStartMs: relative(caller.startedAt),\n callerSpeechOnsetMs: relative(caller.speechOnsetAt),\n callerSpeechOffsetMs: relative(caller.speechOffsetAt),\n callerStreamEndMs: relative(caller.streamEndAt),\n agentFirstPacketMs: relative(ctx.firstPacketAt),\n agentLastPacketMs: relative(ctx.lastPacketAt),\n agentSpeechOnsetMs: relative(ctx.firstAudioAt),\n agentSpeechOffsetMs: relative(ctx.lastAudioAt),\n agentPlaybackEndMs: relative(ctx.playbackEndAt),\n latestTurnBeginMs: relative(ctx.latestTurnBeginAt),\n postTurnBeginFirstPacketMs: relative(ctx.postTurnBeginFirstPacketAt),\n postTurnBeginSpeechOnsetMs: relative(ctx.postTurnBeginFirstAudioAt),\n lastAssistantTranscriptMs: relative(ctx.lastAssistantTranscriptAt),\n callerMaxFrameGapMs: Math.round(caller.maxFrameGapMs ?? 0),\n method: \"client-monotonic-queued-playback-energy-bounds\",\n };\n}\n\nexport async function streamPcmRealtime(omni, pcm, rate, clock, onFrame = () => {}) {\n const frameSize = Math.max(1, Math.round((rate * FRAME_MS) / 1000));\n const bounds = audibleBounds(pcm, rate);\n const startedAt = clock.now();\n let speechOnsetAt = null;\n let speechOffsetAt = null;\n let previousFrameEnd = startedAt;\n let maxFrameGapMs = 0;\n for (let off = 0; off < pcm.length; off += frameSize) {\n const frame = pcm.subarray(off, Math.min(off + frameSize, pcm.length));\n const at = clock.now();\n maxFrameGapMs = Math.max(maxFrameGapMs, at - previousFrameEnd);\n const durationMs = (frame.length / rate) * 1000;\n omni.sendAudio(frame);\n onFrame(frame, at);\n if (bounds.first != null && bounds.first >= off && bounds.first < off + frame.length) {\n speechOnsetAt = at + ((bounds.first - off) / rate) * 1000;\n }\n if (bounds.last != null && bounds.last > off && bounds.last <= off + frame.length) {\n speechOffsetAt = at + ((bounds.last - off) / rate) * 1000;\n }\n previousFrameEnd = at + durationMs;\n // Pace from the actual send. Never burst late frames to \"catch up\".\n await clock.sleep(Math.max(0, previousFrameEnd - clock.now()));\n }\n return { startedAt, speechOnsetAt, speechOffsetAt,\n streamEndAt: clock.now(), maxFrameGapMs };\n}\n\nexport async function streamSilenceWhile(omni, rate, shouldContinue, clock, onFrame = () => {}) {\n const silence = new Int16Array(Math.max(1, Math.round((rate * FRAME_MS) / 1000)));\n while (shouldContinue()) {\n const at = clock.now();\n omni.sendAudio(silence);\n onFrame(silence, at);\n await clock.sleep(FRAME_MS);\n }\n}\n\nexport async function waitForAgentSettle(getCtx, clock, opts = {}) {\n const started = clock.now();\n const observation = opts.callerStreamEndAt != null;\n if (observation && (!Number.isFinite(opts.callerStreamEndAt)\n || opts.callerStreamEndAt < 0 || opts.callerStreamEndAt > started)) {\n throw new Error(\"Invalid caller observation boundary\");\n }\n const timeoutMs = observation ? INTERRUPTION_OBSERVATION.maximumAfterCallerStreamMs : opts.timeoutMs ?? 25000;\n const settleMs = observation ? INTERRUPTION_OBSERVATION.quietMs : opts.settleMs ?? 2000;\n const deadlineAt = (observation ? opts.callerStreamEndAt : started) + timeoutMs;\n const minimumAt = observation ? opts.callerStreamEndAt + INTERRUPTION_OBSERVATION.minimumAfterCallerStreamMs : -Infinity;\n while (true) {\n const ctx = getCtx();\n const at = clock.now();\n if (opts.isClosed?.()) return { reason: \"closed\", at };\n if (observation && at >= deadlineAt) return { reason: \"timeout\", at };\n const hasResponseAudio = !opts.requireTurnBegin || (ctx.latestTurnBeginAt != null\n && ctx.postTurnBeginFirstAudioAt != null);\n // Synthesis text is progress, not a completion event. It can arrive after\n // early/final audio, so extend the quiet floor without demanding another\n // packet after every advisory. The protocol still has no reply-end marker.\n const activityEnd = Math.max(ctx.playbackEndAt ?? -Infinity,\n ctx.latestTurnBeginAt ?? -Infinity, ctx.lastAssistantTranscriptAt ?? -Infinity,\n observation ? ctx.lastCallerTranscriptAt ?? -Infinity : -Infinity);\n // Do not demand a new post-caller turn_begin: an earlier turn can be\n // coalesced and speak later. This records a bounded quiet observation;\n // downstream response-timing checks can impose a stricter begin boundary.\n if (ctx.firstPacketAt != null && hasResponseAudio && at >= minimumAt && at >= activityEnd + settleMs) {\n return { reason: \"settled\", at };\n }\n if (opts.allowEmpty && ctx.firstPacketAt == null && at >= minimumAt && at - started >= (opts.emptyWaitMs ?? 800)) {\n return { reason: \"empty\", at };\n }\n if (at - started >= timeoutMs) return { reason: \"timeout\", at };\n await clock.sleep(observation ? Math.min(20, deadlineAt - at) : 20);\n }\n}\n\nexport async function withTimeout(promise, ms, message, clock) {\n let timer;\n try {\n return await Promise.race([\n promise,\n new Promise((_, reject) => {\n timer = clock.setTimeout(() => reject(new Error(message)), ms);\n }),\n ]);\n } finally {\n clock.clearTimeout(timer);\n }\n}\n",
|
|
6
6
|
"main.mjs": "import PyAI from \"@pyai/sdk\";\nimport WebSocket from \"ws\"; // Explicit transport makes this work on Node 20 too.\nimport { readFile, writeFile, mkdtemp } from \"node:fs/promises\";\nimport { join } from \"node:path\";\nimport { runSession } from \"./session.mjs\";\nimport { readWav, writeWav, assessAnswer } from \"./audio.mjs\";\n\nasync function main() {\n const [callerPath, interruptionPath, extra] = process.argv.slice(2);\n if (!callerPath || extra) throw new Error(\"Usage: node main.mjs caller.wav [interruption.wav]\");\n if (!process.env.PYAI_API_KEY) throw new Error(\"Inject PYAI_API_KEY through your environment first\");\n const caller = readWav(await readFile(callerPath));\n const interruption = interruptionPath ? readWav(await readFile(interruptionPath)) : undefined;\n const pyai = new PyAI({ apiKey: process.env.PYAI_API_KEY, maxRetries: 0,\n fetch: (url, options) => fetch(url, { ...options, signal: AbortSignal.timeout(30000) }),\n });\n const { report, audio, replyAudio } = await runSession({ pyai, webSocket: WebSocket, caller, interruption });\n const directory = await mkdtemp(join(process.cwd(), \"omni-run-\"));\n const save = (name, data) => writeFile(join(directory, name), data, { mode: 0o600, flag: \"wx\" });\n await save(\"session.wav\", writeWav(audio, report.output_rate ?? 24000));\n const reply = writeWav(replyAudio, report.output_rate ?? 24000);\n await save(\"reply.wav\", reply);\n // Hear the captured bytes, rather than trusting synthesis-advisory text.\n let capturedTranscript = \"\";\n if (report.reply_audio_bytes) {\n try {\n const heard = await pyai.audio.transcriptions.create({ file: new File([reply], \"reply.wav\", { type: \"audio/wav\" }), model: \"pyai-hear\" });\n capturedTranscript = heard.text;\n } catch { report.hear_error = \"Captured audio saved; Hear verification failed\"; }\n }\n const result = { ...report, captured_transcript: capturedTranscript, evidence: assessAnswer(report, capturedTranscript) };\n await save(\"report.json\", JSON.stringify(result, null, 2) + \"\\n\");\n console.log(JSON.stringify({ directory, ...result }, null, 2));\n if (result.evidence.captured_answer !== \"verified_by_hear\" || (interruption && result.evidence.interruption !== \"simulated_queue_cleared\")) process.exitCode = 1;\n}\n\nmain().catch(error => {\n const message = String(error.message).replaceAll(process.env.PYAI_API_KEY || \"\\0\", \"[REDACTED]\");\n console.error(JSON.stringify({ error: message, ...(error.omni_code ? { code: error.omni_code } : {}) })); process.exitCode = 1;\n});\n",
|
|
7
7
|
"session.mjs": "import { newAudioCapture, recordAudio, recordTurnBegin, streamPcmRealtime, INTERRUPTION_OBSERVATION } from \"./timing.mjs\";\n\nexport const realClock = {\n now: () => performance.now(),\n sleep: ms => new Promise(resolve => setTimeout(resolve, ms)),\n};\n\n// Only recognized machine codes may leave the example. Error messages, call\n// IDs and arbitrary code values can contain credentials or customer content.\nconst publicErrorCodes = new Set([\n \"invalid_configure\", \"unsupported_tool_transport\", \"tool_configuration_locked\",\n \"media_dead\", \"unauthorized\", \"insufficient_scope\", \"rate_limit_exceeded\",\n \"concurrency_limit_exceeded\", \"daily_cap_exceeded\", \"credit_exhausted\",\n]);\nconst transportErrorCodes = new Set([\"ECONNRESET\", \"ECONNREFUSED\", \"ETIMEDOUT\", \"ENOTFOUND\", \"EAI_AGAIN\"]);\nexport function omniFailure(error) {\n const code = error instanceof Error\n ? (transportErrorCodes.has(error.code) ? error.code : \"transport_or_protocol_error\")\n : (publicErrorCodes.has(error?.code) ? error.code : \"unrecognized_server_error\");\n return Object.assign(new Error(`Omni failed (${code}); no automatic retry was attempted`), { omni_code: code });\n}\n\n/** One bounded, server-side example. Its playback queue is simulated. */\nexport async function runSession({ pyai, webSocket, caller, interruption, clock = realClock, timeoutMs = 60000 }) {\n const rate = 24000, frameSize = 480, silence = new Int16Array(frameSize);\n if (!(caller instanceof Int16Array) || !caller.length || caller.length > rate * 20) throw new Error(\"Provide 1–20 seconds of caller PCM16 at 24 kHz\");\n if (interruption && (!(interruption instanceof Int16Array) || !interruption.length || interruption.length > rate * 20)) throw new Error(\"Invalid interruption PCM\");\n const started = clock.now(), all = newAudioCapture(started), reply = newAudioCapture(started);\n const report = { input_rate: rate, output_rate: null, playback_sink: \"simulated\", tool_calls: 0, tool_executions: 0, duplicate_tool_calls: 0, flushes_after_interruption: 0, cleared_queue_ms: 0, interruption_requested: !!interruption, caller_started_ms: null, interruption_started_ms: null, advisory_transcripts: [], end_reason: \"timeout\" };\n let session, configuredAt = null, closed = false, failure = null, phase = \"greeting\", offset = 0, queueEnd = started, interruptionSent = false, inputEndAt = started;\n const handled = new Set();\n const fail = message => { failure ??= new Error(message); };\n try {\n session = pyai.omni.connect({\n webSocket, rate,\n configure: {\n voice_id: \"stock_amos_en_us\", language: \"en\", greeting: \"Hello. How can I help?\",\n persona: \"When asked about office hours, call lookup_office_hours. Never guess. After the result, say one short sentence giving its opening_time.\",\n tools: [{ name: \"lookup_office_hours\", description: \"Read the synthetic example office opening time.\", side_effect: \"read\", parameters: { type: \"object\", properties: {}, required: [], additionalProperties: false } }],\n },\n onHello: frame => {\n const match = /^pcm16@(8000|24000)$/.exec(frame.audio_out ?? \"\");\n if (!match) return fail(\"Unsupported Omni output format\");\n report.output_rate = Number(match[1]);\n },\n onConfigured: () => { configuredAt ??= clock.now(); },\n onAudio: chunk => {\n if (!report.output_rate) return fail(\"Audio arrived before its output format\");\n const bytes = chunk instanceof ArrayBuffer ? new Uint8Array(chunk) : new Uint8Array(chunk.buffer, chunk.byteOffset, chunk.byteLength);\n if (bytes.byteLength % 2) return fail(\"Invalid PCM16 output\");\n const view = new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength);\n const pcm = Int16Array.from({ length: bytes.length / 2 }, (_, i) => view.getInt16(i * 2, true));\n const now = clock.now();\n recordAudio(all, pcm, now, report.output_rate);\n queueEnd = Math.max(now, queueEnd) + pcm.length / report.output_rate * 1000;\n if (report.tool_executions) recordAudio(reply, pcm, now, report.output_rate);\n },\n onEvent: frame => {\n if (frame.event === \"turn_begin\") recordTurnBegin(all, clock.now(), frame.turn);\n },\n onTranscript: frame => {\n // An assistant advisory is progress; never treat it as received speech.\n if (report.advisory_transcripts.length < 100) report.advisory_transcripts.push({ role: frame.role, text: frame.text });\n if (frame.role === \"assistant\") all.lastAssistantTranscriptAt = clock.now();\n },\n onToolCall: frame => {\n report.tool_calls++;\n if (handled.has(frame.call_id)) { report.duplicate_tool_calls++; return; }\n handled.add(frame.call_id);\n if (frame.name !== \"lookup_office_hours\") { session.toolResult(frame.call_id, { error: \"Unknown example tool\" }); return; }\n // Replace this read-only fixture with your authorized data lookup.\n report.tool_executions++;\n session.toolResult(frame.call_id, { result: { opening_time: \"9 a.m.\", source: \"synthetic example office\" } });\n },\n onBargeIn: () => {\n const now = clock.now(), queued = Math.max(0, queueEnd - now);\n if (report.interruption_started_ms != null) { report.flushes_after_interruption++; report.cleared_queue_ms += queued; }\n // A real speaker adapter must cancel scheduled and currently playing audio.\n queueEnd = now; all.playbackEndAt = now; reply.playbackEndAt = now;\n },\n onError: error => { failure ??= omniFailure(error); },\n onClose: () => { closed = true; },\n });\n while (clock.now() - started < timeoutMs) {\n if (failure) throw failure;\n if (closed) { report.end_reason = \"closed\"; break; }\n const now = clock.now();\n if (configuredAt == null || !report.output_rate) {\n if (now - started >= 10000) throw new Error(\"Omni connection/configuration timed out\");\n await clock.sleep(20); continue;\n }\n const quietAt = Math.max(queueEnd, all.lastPacketAt ?? now, all.lastAssistantTranscriptAt ?? 0, inputEndAt);\n if (phase === \"greeting\") {\n if (all.firstPacketAt != null && now - configuredAt >= 2000 && now >= quietAt + 800) {\n phase = \"caller\"; report.caller_started_ms = now - started;\n } else if (now - configuredAt > 15000) throw new Error(\"Greeting did not drain within 15 seconds\");\n }\n if (phase === \"reply\" && interruption && !interruptionSent && reply.samples && queueEnd - now >= 200) {\n phase = \"interruption\"; offset = 0; interruptionSent = true; report.interruption_started_ms = now - started;\n }\n const observationEnd = inputEndAt + (interruptionSent ? INTERRUPTION_OBSERVATION.minimumAfterCallerStreamMs : 0);\n if (phase === \"reply\" && reply.samples && now >= quietAt + 2000 && now >= observationEnd && (!interruption || interruptionSent)) {\n report.end_reason = \"quiet_window\"; break;\n }\n // Exactly one producer: each 20 ms slot contains caller PCM OR silence.\n let pcm = silence;\n if (phase === \"caller\" || phase === \"interruption\") {\n const source = phase === \"caller\" ? caller : interruption;\n pcm = source.subarray(offset, offset + frameSize); offset += pcm.length;\n if (offset >= source.length) {\n inputEndAt = now + pcm.length / rate * 1000;\n report[phase === \"caller\" ? \"caller_completed_ms\" : \"interruption_completed_ms\"] = inputEndAt - started;\n phase = \"reply\"; offset = 0;\n }\n }\n await streamPcmRealtime(session, pcm, rate, clock);\n }\n report.audio_bytes = all.samples * 2;\n report.reply_audio_bytes = reply.samples * 2;\n report.elapsed_ms = Math.round(clock.now() - started);\n return { report, audio: all.pcm, replyAudio: reply.pcm };\n } finally { session?.close(); }\n}\n",
|
|
@@ -109,7 +109,7 @@ export class PyaiDialerElement extends HTMLElement {
|
|
|
109
109
|
complete.hidden = transfer?.mode !== "warm" || transfer.status !== "ready";
|
|
110
110
|
complete.disabled = Boolean(snapshot.transferPending);
|
|
111
111
|
const cancel = this.root.querySelector("[data-action=cancel-transfer]");
|
|
112
|
-
cancel.hidden = transfer
|
|
112
|
+
cancel.hidden = !transfer || !pending;
|
|
113
113
|
cancel.disabled = Boolean(snapshot.transferPending || ["completing", "canceling"].includes(transfer?.status ?? ""));
|
|
114
114
|
this.root.querySelector("[data-action=recover-transfer]").hidden = !snapshot.transferPending;
|
|
115
115
|
}
|
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import type { TelephonyBrowserSession, TelephonyBrowserSessionCreate, TelephonyCall } from "../index.ts";
|
|
1
|
+
import type { TelephonyBrowserSession, TelephonyBrowserSessionCreate, TelephonyCall, TelephonyCallControl } from "../index.ts";
|
|
2
2
|
/** Implement on your server using PyAI's telephony client. The browser receives
|
|
3
3
|
* a single-call capability, never an organization key or carrier credential. */
|
|
4
4
|
export interface DialerBroker {
|
|
@@ -6,6 +6,7 @@ export interface DialerBroker {
|
|
|
6
6
|
get(callId: string): Promise<TelephonyCall>;
|
|
7
7
|
hangup(callId: string): Promise<TelephonyCall>;
|
|
8
8
|
setHold?(callId: string, held: boolean, key: string): Promise<TelephonyCall>;
|
|
9
|
+
sendDtmf?(callId: string, digits: string, key: string): Promise<TelephonyCallControl>;
|
|
9
10
|
transfers?: {
|
|
10
11
|
create(callId: string, input: {
|
|
11
12
|
to: string;
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
const terminal = (call) => ["completed", "failed", "canceled"].includes(call.status);
|
|
2
2
|
/** A durable handoff keeps its original room after the browser goes away. */
|
|
3
|
-
export const dialerCallHandedOff = (call) => call.transfer_enabled === true && Boolean(call.transfers?.some(transfer => transfer.status === "completed" || transfer.mode === "cold" &&
|
|
3
|
+
export const dialerCallHandedOff = (call) => call.transfer_enabled === true && Boolean(call.transfers?.some(transfer => transfer.status === "completed" || transfer.mode === "cold" && transfer.status === "completing"));
|
|
4
4
|
/** One immutable workspace/user scope per instance. Setup and stop retain the
|
|
5
5
|
* original intent through lost acknowledgments, reloads and audio failures. */
|
|
6
6
|
export class ManagedDialer {
|
|
@@ -291,8 +291,10 @@ export class ManagedDialer {
|
|
|
291
291
|
const transfer = this.snapshot.call?.transfers?.find(item => item.id === this.intent.transfer.id);
|
|
292
292
|
if (command === "complete" && transfer?.status !== "ready")
|
|
293
293
|
throw new Error("Wait until the transfer target has joined.");
|
|
294
|
-
if (this.intent.transfer.mode !== "warm")
|
|
294
|
+
if (this.intent.transfer.mode !== "warm" && command === "complete")
|
|
295
295
|
throw new Error("Cold transfers complete automatically.");
|
|
296
|
+
if (command === "cancel" && ["completing", "completed"].includes(transfer?.status ?? ""))
|
|
297
|
+
throw new Error("The handoff has started. Check its status instead of canceling it.");
|
|
296
298
|
if (this.intent.transfer.command && this.intent.transfer.command !== command)
|
|
297
299
|
throw new Error("Recover the saved transfer command first.");
|
|
298
300
|
const next = structuredClone(this.intent);
|
|
@@ -1,6 +1,5 @@
|
|
|
1
1
|
/** Server-only broker. Never import this module into a browser bundle. */
|
|
2
2
|
import { createHash } from "node:crypto";
|
|
3
|
-
import { dialerCallHandedOff } from "./managedDialer.js";
|
|
4
3
|
export function createServerDialerBroker(sdk, authenticatedUserId) {
|
|
5
4
|
if (!authenticatedUserId.trim())
|
|
6
5
|
throw new Error("An authenticated user is required.");
|
|
@@ -20,10 +19,9 @@ export function createServerDialerBroker(sdk, authenticatedUserId) {
|
|
|
20
19
|
},
|
|
21
20
|
get: ownedCall,
|
|
22
21
|
async hangup(id) {
|
|
23
|
-
|
|
24
|
-
|
|
25
|
-
|
|
26
|
-
return sdk.telephony.calls.hangup(id);
|
|
22
|
+
await ownedCall(id);
|
|
23
|
+
// The provider rechecks handoff ownership in the stop CAS, after this read.
|
|
24
|
+
return sdk.telephony.calls.hangup(id, { user_id: authenticatedUserId });
|
|
27
25
|
},
|
|
28
26
|
async setHold(id, held, key) {
|
|
29
27
|
await ownedCall(id);
|
|
@@ -32,6 +30,13 @@ export function createServerDialerBroker(sdk, authenticatedUserId) {
|
|
|
32
30
|
const scopedKey = "dialer-hold:" + createHash("sha256").update(authenticatedUserId).update("\0").update(id).update("\0").update(key).digest("hex");
|
|
33
31
|
return sdk.telephony.calls.setHold(id, { held, user_id: authenticatedUserId }, { idempotencyKey: scopedKey });
|
|
34
32
|
},
|
|
33
|
+
async sendDtmf(id, digits, key) {
|
|
34
|
+
await ownedCall(id);
|
|
35
|
+
if (!key || key.length > 255 || /[\r\n]/.test(key))
|
|
36
|
+
throw new Error("A valid keypad retry key is required.");
|
|
37
|
+
const scopedKey = "dialer-keypad:" + createHash("sha256").update(authenticatedUserId).update("\0").update(id).update("\0").update(key).digest("hex");
|
|
38
|
+
return sdk.telephony.calls.sendDtmf(id, { digits, user_id: authenticatedUserId }, { idempotencyKey: scopedKey });
|
|
39
|
+
},
|
|
35
40
|
transfers: {
|
|
36
41
|
async create(id, input, key) {
|
|
37
42
|
await ownedCall(id);
|
package/dist/index.d.ts
CHANGED
|
@@ -860,6 +860,15 @@ export interface TelephonyCall {
|
|
|
860
860
|
held?: boolean;
|
|
861
861
|
hold_pending?: boolean;
|
|
862
862
|
transfers?: TelephonyTransfer[];
|
|
863
|
+
forwarding?: {
|
|
864
|
+
to: string | null;
|
|
865
|
+
status: "waiting" | "claimed" | "unknown" | "acknowledged" | "canceled";
|
|
866
|
+
menu?: {
|
|
867
|
+
kind: "keypad";
|
|
868
|
+
status: "pending" | "selected" | "rejected";
|
|
869
|
+
choice_id: string | null;
|
|
870
|
+
};
|
|
871
|
+
};
|
|
863
872
|
user_id?: string;
|
|
864
873
|
client_reference: string | null;
|
|
865
874
|
stop_requested: boolean;
|
|
@@ -907,6 +916,13 @@ export interface TelephonyCallCapabilities {
|
|
|
907
916
|
};
|
|
908
917
|
revalidation_required: true;
|
|
909
918
|
}
|
|
919
|
+
export interface TelephonyCallControl {
|
|
920
|
+
object: "telephony.call_control";
|
|
921
|
+
id: string;
|
|
922
|
+
call_id: string;
|
|
923
|
+
action: "send_dtmf";
|
|
924
|
+
status: "acknowledged" | "unknown" | "rejected";
|
|
925
|
+
}
|
|
910
926
|
export interface TelephonyCallCreate {
|
|
911
927
|
from_number_id: string;
|
|
912
928
|
to: string;
|
|
@@ -958,6 +974,15 @@ export interface TelephonyAvailableNumber {
|
|
|
958
974
|
region?: string | null;
|
|
959
975
|
capabilities?: TelephonyCapabilities;
|
|
960
976
|
monthly_cost_cents?: number;
|
|
977
|
+
/** Bind purchase to the exact account/market inventory result. */
|
|
978
|
+
inventory_reference?: string;
|
|
979
|
+
}
|
|
980
|
+
export interface TelephonyNumberInventory extends ListResponse<TelephonyAvailableNumber> {
|
|
981
|
+
provisioning?: {
|
|
982
|
+
mode: "automated" | "assisted";
|
|
983
|
+
capability: string;
|
|
984
|
+
message: string;
|
|
985
|
+
};
|
|
961
986
|
}
|
|
962
987
|
export interface TelephonyNumber {
|
|
963
988
|
object?: "telephony.number";
|
|
@@ -980,7 +1005,7 @@ export interface TelephonyNumberOrder {
|
|
|
980
1005
|
id: string;
|
|
981
1006
|
status: "pending" | "unknown" | "completed" | "failed" | "released";
|
|
982
1007
|
phone_number: string;
|
|
983
|
-
country: "AU";
|
|
1008
|
+
country: "AU" | "US" | "CA";
|
|
984
1009
|
project_id: string;
|
|
985
1010
|
retry_after_seconds: number;
|
|
986
1011
|
message: string;
|
|
@@ -989,12 +1014,72 @@ export interface TelephonyNumberOrder {
|
|
|
989
1014
|
}
|
|
990
1015
|
export interface TelephonyNumberPurchase {
|
|
991
1016
|
phone_number: string;
|
|
1017
|
+
country?: "US" | "CA" | "IN" | "AU";
|
|
1018
|
+
inventory_reference?: string;
|
|
992
1019
|
agent_id?: string | null;
|
|
993
1020
|
provisioning_mode?: "automated" | "adopt_preowned";
|
|
994
1021
|
project_id?: string;
|
|
995
1022
|
compliance_case_id?: string;
|
|
996
1023
|
max_monthly_cost_cents?: number;
|
|
997
1024
|
}
|
|
1025
|
+
export interface TelephonyInboundBinding {
|
|
1026
|
+
object: "telephony.inbound_binding";
|
|
1027
|
+
number_id: string;
|
|
1028
|
+
version: number;
|
|
1029
|
+
enabled: boolean;
|
|
1030
|
+
project_id?: string;
|
|
1031
|
+
agent_id?: string;
|
|
1032
|
+
agent_release_id?: string;
|
|
1033
|
+
agent_profile_sha256?: string;
|
|
1034
|
+
admission_mode?: "client" | "platform";
|
|
1035
|
+
forward?: TelephonyInboundForward | null;
|
|
1036
|
+
carrier_operation?: {
|
|
1037
|
+
id: string;
|
|
1038
|
+
phase: "pending" | "unknown" | "active" | "rollback_pending" | "rolled_back";
|
|
1039
|
+
action: "activate" | "rollback";
|
|
1040
|
+
};
|
|
1041
|
+
}
|
|
1042
|
+
export interface TelephonyInboundMenu {
|
|
1043
|
+
/** Saved single-digit spoken menu before forwarding. */
|
|
1044
|
+
kind: "keypad";
|
|
1045
|
+
prompt: string;
|
|
1046
|
+
choices: Array<{
|
|
1047
|
+
id: string;
|
|
1048
|
+
key: string;
|
|
1049
|
+
to: string;
|
|
1050
|
+
}>;
|
|
1051
|
+
timeout_seconds: number;
|
|
1052
|
+
/** null hangs up; otherwise the ID of one saved choice. */
|
|
1053
|
+
timeout_choice_id: string | null;
|
|
1054
|
+
}
|
|
1055
|
+
export type TelephonyInboundForward = {
|
|
1056
|
+
max_duration_seconds: number;
|
|
1057
|
+
recording: boolean;
|
|
1058
|
+
} & ({
|
|
1059
|
+
to: string;
|
|
1060
|
+
menu?: never;
|
|
1061
|
+
} | {
|
|
1062
|
+
to?: never;
|
|
1063
|
+
menu: TelephonyInboundMenu;
|
|
1064
|
+
});
|
|
1065
|
+
export type TelephonyInboundBindingUpdate = {
|
|
1066
|
+
expected_version: number;
|
|
1067
|
+
activate_carrier_route?: boolean;
|
|
1068
|
+
rollback_carrier_route?: boolean;
|
|
1069
|
+
/** Persist the original operation ID; recover it instead of inventing a new one. */
|
|
1070
|
+
carrier_operation_id?: string;
|
|
1071
|
+
} & ({
|
|
1072
|
+
enabled: false;
|
|
1073
|
+
} | {
|
|
1074
|
+
enabled: true;
|
|
1075
|
+
project_id: string;
|
|
1076
|
+
agent_id: string;
|
|
1077
|
+
agent_release_id: string;
|
|
1078
|
+
agent_profile_sha256: string;
|
|
1079
|
+
admission_mode?: "client" | "platform";
|
|
1080
|
+
agent_knowledge_receipt?: Record<string, unknown>;
|
|
1081
|
+
forward?: TelephonyInboundForward | null;
|
|
1082
|
+
});
|
|
998
1083
|
export type TraceVerdict = "PASS" | "WARN" | "FAIL";
|
|
999
1084
|
export type TraceSeverity = "low" | "medium" | "high" | "critical";
|
|
1000
1085
|
export interface TraceInteraction {
|
|
@@ -1551,7 +1636,9 @@ export declare class PyAI {
|
|
|
1551
1636
|
next_cursor: string | null;
|
|
1552
1637
|
}>;
|
|
1553
1638
|
/** Durable stop intent; use get() until a terminal status is confirmed. */
|
|
1554
|
-
hangup: (id: string
|
|
1639
|
+
hangup: (id: string, input?: {
|
|
1640
|
+
user_id: string;
|
|
1641
|
+
}) => Promise<TelephonyCall>;
|
|
1555
1642
|
decideAdmission: (id: string, input: {
|
|
1556
1643
|
decision: "accept" | "reject";
|
|
1557
1644
|
client_reference: string;
|
|
@@ -1568,6 +1655,14 @@ export declare class PyAI {
|
|
|
1568
1655
|
}, options: {
|
|
1569
1656
|
idempotencyKey: string;
|
|
1570
1657
|
}) => Promise<TelephonyCall>;
|
|
1658
|
+
/** Server-side keypad command for an owned switch call. Reuse the same key
|
|
1659
|
+
* after an unknown result; never store keypad digits in a browser journal. */
|
|
1660
|
+
sendDtmf: (id: string, input: {
|
|
1661
|
+
digits: string;
|
|
1662
|
+
user_id: string;
|
|
1663
|
+
}, options: {
|
|
1664
|
+
idempotencyKey: string;
|
|
1665
|
+
}) => Promise<TelephonyCallControl>;
|
|
1571
1666
|
transfers: {
|
|
1572
1667
|
/** Persist the key and target before requesting a consultation. */
|
|
1573
1668
|
create: (id: string, input: {
|
|
@@ -1594,7 +1689,7 @@ export declare class PyAI {
|
|
|
1594
1689
|
areaCode?: string;
|
|
1595
1690
|
contains?: string;
|
|
1596
1691
|
limit?: number;
|
|
1597
|
-
}) => Promise<
|
|
1692
|
+
}) => Promise<TelephonyNumberInventory>;
|
|
1598
1693
|
/** List the org's managed numbers (active only unless `includeReleased`). */
|
|
1599
1694
|
list: (params?: {
|
|
1600
1695
|
includeReleased?: boolean;
|
|
@@ -1605,6 +1700,10 @@ export declare class PyAI {
|
|
|
1605
1700
|
}) => Promise<TelephonyNumber | TelephonyNumberOrder>;
|
|
1606
1701
|
/** Route a number to an agent (`agentId: null` to unassign). */
|
|
1607
1702
|
assign: (id: string, agentId: string | null) => Promise<TelephonyNumber>;
|
|
1703
|
+
getInboundBinding: (id: string) => Promise<TelephonyInboundBinding>;
|
|
1704
|
+
/** Save the profile and stable operation first; after uncertainty read the
|
|
1705
|
+
* binding before explicitly recovering that same operation. */
|
|
1706
|
+
setInboundBinding: (id: string, input: TelephonyInboundBindingUpdate) => Promise<TelephonyInboundBinding>;
|
|
1608
1707
|
/** Release a number back to the carrier (idempotent). */
|
|
1609
1708
|
release: (id: string) => Promise<TelephonyNumber>;
|
|
1610
1709
|
};
|
package/dist/index.js
CHANGED
|
@@ -1193,12 +1193,19 @@ export class PyAI {
|
|
|
1193
1193
|
return this.getJson(`/v1/telephony/calls${query.size ? `?${query}` : ""}`);
|
|
1194
1194
|
},
|
|
1195
1195
|
/** Durable stop intent; use get() until a terminal status is confirmed. */
|
|
1196
|
-
hangup: (id) => this.postJson(`/v1/telephony/calls/${encodeURIComponent(id)}/hangup`, {}),
|
|
1196
|
+
hangup: (id, input) => this.postJson(`/v1/telephony/calls/${encodeURIComponent(id)}/hangup`, input ?? {}),
|
|
1197
1197
|
decideAdmission: (id, input) => this.postJson(`/v1/telephony/calls/${encodeURIComponent(id)}/admission`, input),
|
|
1198
1198
|
recording: (id, options = {}) => this.request(`/v1/telephony/calls/${encodeURIComponent(id)}/recording`, { headers: this.authHeaders(options.range ? { Range: options.range } : {}) }),
|
|
1199
1199
|
deleteRecording: async (id) => { await this.deleteReq(`/v1/telephony/calls/${encodeURIComponent(id)}/recording`); },
|
|
1200
1200
|
/** The same durable key recovers an uncertain hold command without repeating it. */
|
|
1201
1201
|
setHold: (id, input, options) => this.postJson(`/v1/telephony/calls/${encodeURIComponent(id)}/hold`, input, telephonyIdempotencyHeader(options.idempotencyKey)),
|
|
1202
|
+
/** Server-side keypad command for an owned switch call. Reuse the same key
|
|
1203
|
+
* after an unknown result; never store keypad digits in a browser journal. */
|
|
1204
|
+
sendDtmf: (id, input, options) => {
|
|
1205
|
+
if (!/^[0-9A-D*#]{1,32}$/.test(input.digits))
|
|
1206
|
+
throw new Error("Use 1–32 keypad digits (0–9, *, #, A–D).");
|
|
1207
|
+
return this.postJson(`/v1/telephony/calls/${encodeURIComponent(id)}/dtmf`, input, telephonyIdempotencyHeader(options.idempotencyKey));
|
|
1208
|
+
},
|
|
1202
1209
|
transfers: {
|
|
1203
1210
|
/** Persist the key and target before requesting a consultation. */
|
|
1204
1211
|
create: (id, input, options) => this.postJson(`/v1/telephony/calls/${encodeURIComponent(id)}/transfers`, input, telephonyIdempotencyHeader(options.idempotencyKey)),
|
|
@@ -1234,6 +1241,10 @@ export class PyAI {
|
|
|
1234
1241
|
buy: (params, options) => this.postJson("/v1/telephony/numbers", params, telephonyIdempotencyHeader(options.idempotencyKey)),
|
|
1235
1242
|
/** Route a number to an agent (`agentId: null` to unassign). */
|
|
1236
1243
|
assign: (id, agentId) => this.postJson(`/v1/telephony/numbers/${encodeURIComponent(id)}/assign`, { agent_id: agentId }),
|
|
1244
|
+
getInboundBinding: (id) => this.getJson(`/v1/telephony/numbers/${encodeURIComponent(id)}/inbound-binding`),
|
|
1245
|
+
/** Save the profile and stable operation first; after uncertainty read the
|
|
1246
|
+
* binding before explicitly recovering that same operation. */
|
|
1247
|
+
setInboundBinding: (id, input) => this.putJson(`/v1/telephony/numbers/${encodeURIComponent(id)}/inbound-binding`, input),
|
|
1237
1248
|
/** Release a number back to the carrier (idempotent). */
|
|
1238
1249
|
release: async (id) => {
|
|
1239
1250
|
const res = await this.deleteReq(`/v1/telephony/numbers/${encodeURIComponent(id)}`);
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@pyai/sdk",
|
|
3
|
-
"version": "0.
|
|
3
|
+
"version": "0.17.0",
|
|
4
4
|
"description": "Official TypeScript/JavaScript SDK for PyAI, speech-to-text (Hear), text-to-speech (Speak), realtime voice agents (Omni), and call compliance (Trace).",
|
|
5
5
|
"type": "module",
|
|
6
6
|
"main": "dist/index.js",
|
package/src/cli-omni-template.ts
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
// Generated by scripts/generate-omni-starter.mjs; edit examples/omni-starter instead.
|
|
2
2
|
export const omniTemplate = Object.freeze({
|
|
3
|
-
"README.md": "# Runnable Omni starter\n\nRequires Node 20.19+ and ESM. Node 20 uses the included `ws` transport;\nNode 22 also works. The separate MCP server requires Node 22+.\nThis server-side example uses a synthetic read-only office-hours tool. Read\n`PYAI.md` when generating the project through the CLI.\n\n<!-- omni-install:start -->\n```sh\nnpm install @pyai/sdk@0.
|
|
4
|
-
"package.json": "{\n \"name\": \"pyai-omni-starter\",\n \"version\": \"0.0.0\",\n \"private\": true,\n \"type\": \"module\",\n \"engines\": {\n \"node\": \">=20.19.0\"\n },\n \"scripts\": {\n \"start\": \"node main.mjs\"\n },\n \"dependencies\": {\n \"@pyai/sdk\": \"0.
|
|
3
|
+
"README.md": "# Runnable Omni starter\n\nRequires Node 20.19+ and ESM. Node 20 uses the included `ws` transport;\nNode 22 also works. The separate MCP server requires Node 22+.\nThis server-side example uses a synthetic read-only office-hours tool. Read\n`PYAI.md` when generating the project through the CLI.\n\n<!-- omni-install:start -->\n```sh\nnpm install @pyai/sdk@0.17.0\nnpx pyai init voice-demo --template omni\ncd voice-demo\nnpm install\n```\n\nInitialization is offline; installing dependencies and running `main.mjs` are\nseparate steps. For an existing checkout of this example, run `npm install`\nin its directory instead.\n<!-- omni-install:end -->\n\nInject `PYAI_API_KEY` through your secret manager or shell environment. Do not\npaste a real key into source, a prompt, or shell history. `.env.example` documents\nthe variable; this program does not load `.env` or saved CLI profiles.\n\nPrepare a 24 kHz PCM16 mono WAV, at most 20 seconds, saying “Please look up the\noffice opening time.” Then run:\n\n```sh\nnode main.mjs caller.wav\n```\n\nRunning the program explicitly opens one bounded Omni session and uploads its\ncaptured reply to Hear. Both consume your key's allowed usage. Automatic retries\nare disabled. No managed number is bought and no phone call is placed.\n\n`session.mjs` waits for configuration and the greeting to drain, sends exactly\none paced input stream (caller PCM or silence), returns the read tool's result,\nand captures output using the rate from `hello.audio_out`. Never run a separate\nsilence timer alongside caller audio. `timing.mjs` is generated from the existing\nOmni evaluation harness's timing utilities.\n\nThe new private `omni-run-*` folder contains `session.wav`, `reply.wav`, and\n`report.json`. The report separates received audio, an office-hours answer\nrecovered by Hear from captured audio, and physical playback (not tested).\nThe office-hours check accepts formatting such as `opens@9a.m`, `9 a.m.` and\n`nine in the morning` in a complete affirmative opening-time sentence. Wrong\ntimes, negation, conflicting or truncated text remain unverified. This is a\nconservative check for the example fact, not a general answer-quality score.\nA synthesis transcript is advisory; it is not evidence that speech arrived.\nThis starter does not negotiate playback receipts. Capture ends after the\nsimulated queue drains and two seconds of quiet, a bounded capture heuristic.\nListen to the saved file when qualifying actual sound quality or completeness.\n\nTo exercise interruption, supply a second WAV saying “Stop speaking now”:\n\n```sh\nnode main.mjs caller.wav interruption.wav\n```\n\nThe second clip starts while reply audio is queued. `onBargeIn` clears that\nsimulated queue, and capture observes at least five seconds after that clip ends.\nAn interrupted answer can correctly fail the full-answer check;\ninspect the separate interruption evidence. This is not a physical speaker test.\nFor a real playback adapter, cancel both queued and currently playing audio.\nReplace the example lookup with authorized application data before product use.\n\nOn a session error, the program exits with JSON containing a recognized machine\n`code` (for example `media_dead` or `invalid_configure`). Unknown server codes\nbecome `unrecognized_server_error`; transport or decoding errors without a\nrecognized transport code become `transport_or_protocol_error`. Raw error text,\ncall IDs and unknown code values are omitted. No automatic retry is attempted;\ninspect the code and your authorized diagnostics before deciding to retry.\n",
|
|
4
|
+
"package.json": "{\n \"name\": \"pyai-omni-starter\",\n \"version\": \"0.0.0\",\n \"private\": true,\n \"type\": \"module\",\n \"engines\": {\n \"node\": \">=20.19.0\"\n },\n \"scripts\": {\n \"start\": \"node main.mjs\"\n },\n \"dependencies\": {\n \"@pyai/sdk\": \"0.17.0\",\n \"ws\": \"^8.18.3\"\n }\n}\n",
|
|
5
5
|
"timing.mjs": "// Monotonic capture timing. Arrival time is distinct from queued playback:\n// a single packet may contain seconds of audio and must finish before settling.\nexport const FRAME_MS = 20;\nexport const REALTIME_GAP_LIMIT_MS = 100;\n// Interruption tests need an observation window after caller transmission,\n// even when an earlier cue/response already finished. These are test bounds,\n// not an inferred engine reply-end or a naturalness threshold.\nexport const INTERRUPTION_OBSERVATION = Object.freeze({\n minimumAfterCallerStreamMs: 5000, quietMs: 2000, maximumAfterCallerStreamMs: 25000,\n});\n\n/** Approximate audible bounds; this is energy detection, not a speech/VAD claim. */\nexport function audibleBounds(pcm, rate, rmsFloor = 160) {\n const window = Math.max(1, Math.round(rate / 100)); // 10 ms windows\n let first = null;\n let last = null;\n for (let off = 0; off < pcm.length; off += window) {\n const end = Math.min(pcm.length, off + window);\n let energy = 0;\n for (let i = off; i < end; i++) energy += pcm[i] * pcm[i];\n if (Math.sqrt(energy / (end - off)) >= rmsFloor) {\n first ??= off;\n last = end;\n }\n }\n return { first, last };\n}\n\nexport function newAudioCapture(started) {\n return {\n started, firstPacketAt: null, lastPacketAt: null,\n firstAudioAt: null, lastAudioAt: null, playbackEndAt: null,\n samples: 0, pcm: [], turnBeginAt: null, eouMs: null,\n latestTurnBeginAt: null, postTurnBeginFirstPacketAt: null,\n postTurnBeginFirstAudioAt: null, lastAssistantTranscriptAt: null, lastCallerTranscriptAt: null,\n callerTranscriptEvents: [], assistantTranscriptEvents: [],\n turnBegins: [], unheardCancellations: [], unheardSupersededTurns: [], tools: [], toolResults: [], kb: null, events: [],\n };\n}\n\nexport function recordTurnBegin(ctx, at, turn, origin = 0) {\n const prior = ctx.turnBegins.at(-1);\n const cancellation = ctx.unheardCancellations.at(-1);\n if (prior && cancellation && !cancellation.admitted\n && cancellation.turn === prior.turn && cancellation.turnBeginIndex === ctx.turnBegins.length - 1\n && Number.isSafeInteger(turn) && turn > prior.turn && at > cancellation.at\n && ctx.samples === 0 && ctx.assistantTranscriptEvents.length === 0\n && ctx.tools.length === 0 && ctx.toolResults.length === 0) {\n cancellation.admitted = true;\n ctx.unheardSupersededTurns.push({ turn: prior.turn, beginAtMs: prior.atMs,\n cancelledAtMs: cancellation.atMs, cancellationEventIndex: cancellation.eventIndex,\n replacementTurn: turn, outputPacketsBeforeReplacement: 0 });\n }\n ctx.turnBeginAt ??= at;\n ctx.latestTurnBeginAt = at;\n // Earlier PCM can be a listening cue. A new reply is still pending even\n // when that cue has already finished playing; retain all of its PCM.\n ctx.postTurnBeginFirstPacketAt = null;\n ctx.postTurnBeginFirstAudioAt = null;\n ctx.turnBegins.push({ atMs: Math.round(at - origin), turn: turn ?? null });\n}\n\nexport function recordUnheardCancellation(ctx, event, at, origin, eventIndex) {\n const prior = ctx.turnBegins.at(-1);\n if (event.event !== \"flush\" || event.reason !== \"turn_merge\"\n || !Number.isSafeInteger(event.cancelled_turn) || event.cancelled_turn <= 0\n || prior?.turn !== event.cancelled_turn || at < ctx.latestTurnBeginAt\n || ctx.samples !== 0 || ctx.assistantTranscriptEvents.length\n || ctx.tools.length || ctx.toolResults.length\n || ctx.unheardCancellations.some(c => c.turnBeginIndex === ctx.turnBegins.length - 1)) return;\n ctx.unheardCancellations.push({ turn: event.cancelled_turn, at,\n atMs: Math.round(at - origin), eventIndex, turnBeginIndex: ctx.turnBegins.length - 1,\n admitted: false });\n}\n\nexport function recordAudio(ctx, pcm, at, rate) {\n if (!(pcm instanceof Int16Array) || !pcm.length) return null;\n const copy = pcm.slice(); // Some transports reuse callback buffers.\n const playbackAt = Math.max(at, ctx.playbackEndAt ?? at);\n const endAt = playbackAt + (copy.length / rate) * 1000;\n const bounds = audibleBounds(copy, rate);\n ctx.firstPacketAt ??= at;\n ctx.lastPacketAt = at;\n ctx.playbackEndAt = endAt;\n const afterTurnBegin = ctx.latestTurnBeginAt != null && at >= ctx.latestTurnBeginAt;\n if (afterTurnBegin) ctx.postTurnBeginFirstPacketAt ??= at;\n if (bounds.first != null) {\n ctx.firstAudioAt ??= playbackAt + (bounds.first / rate) * 1000;\n ctx.lastAudioAt = playbackAt + (bounds.last / rate) * 1000;\n if (afterTurnBegin) ctx.postTurnBeginFirstAudioAt ??= playbackAt + (bounds.first / rate) * 1000;\n }\n ctx.samples += copy.length;\n ctx.pcm.push(copy);\n return { playbackAt, endAt, pcm: copy };\n}\n\n/** Preserve negative gaps: they are overlaps, never missing measurements. */\nexport function captureTiming(ctx, caller, origin = 0) {\n const relative = (at) => at == null ? null : Math.round(at - origin);\n const gap = (at) => at == null || caller.speechOffsetAt == null\n ? null : Math.round(at - caller.speechOffsetAt);\n return {\n ttfbMs: gap(ctx.firstAudioAt),\n anyAudioTtfbMs: gap(ctx.firstAudioAt),\n postTurnBeginTtfbMs: gap(ctx.postTurnBeginFirstAudioAt),\n turnMs: gap(ctx.lastAudioAt),\n packetTtfbMs: gap(ctx.firstPacketAt),\n callerStartMs: relative(caller.startedAt),\n callerSpeechOnsetMs: relative(caller.speechOnsetAt),\n callerSpeechOffsetMs: relative(caller.speechOffsetAt),\n callerStreamEndMs: relative(caller.streamEndAt),\n agentFirstPacketMs: relative(ctx.firstPacketAt),\n agentLastPacketMs: relative(ctx.lastPacketAt),\n agentSpeechOnsetMs: relative(ctx.firstAudioAt),\n agentSpeechOffsetMs: relative(ctx.lastAudioAt),\n agentPlaybackEndMs: relative(ctx.playbackEndAt),\n latestTurnBeginMs: relative(ctx.latestTurnBeginAt),\n postTurnBeginFirstPacketMs: relative(ctx.postTurnBeginFirstPacketAt),\n postTurnBeginSpeechOnsetMs: relative(ctx.postTurnBeginFirstAudioAt),\n lastAssistantTranscriptMs: relative(ctx.lastAssistantTranscriptAt),\n callerMaxFrameGapMs: Math.round(caller.maxFrameGapMs ?? 0),\n method: \"client-monotonic-queued-playback-energy-bounds\",\n };\n}\n\nexport async function streamPcmRealtime(omni, pcm, rate, clock, onFrame = () => {}) {\n const frameSize = Math.max(1, Math.round((rate * FRAME_MS) / 1000));\n const bounds = audibleBounds(pcm, rate);\n const startedAt = clock.now();\n let speechOnsetAt = null;\n let speechOffsetAt = null;\n let previousFrameEnd = startedAt;\n let maxFrameGapMs = 0;\n for (let off = 0; off < pcm.length; off += frameSize) {\n const frame = pcm.subarray(off, Math.min(off + frameSize, pcm.length));\n const at = clock.now();\n maxFrameGapMs = Math.max(maxFrameGapMs, at - previousFrameEnd);\n const durationMs = (frame.length / rate) * 1000;\n omni.sendAudio(frame);\n onFrame(frame, at);\n if (bounds.first != null && bounds.first >= off && bounds.first < off + frame.length) {\n speechOnsetAt = at + ((bounds.first - off) / rate) * 1000;\n }\n if (bounds.last != null && bounds.last > off && bounds.last <= off + frame.length) {\n speechOffsetAt = at + ((bounds.last - off) / rate) * 1000;\n }\n previousFrameEnd = at + durationMs;\n // Pace from the actual send. Never burst late frames to \"catch up\".\n await clock.sleep(Math.max(0, previousFrameEnd - clock.now()));\n }\n return { startedAt, speechOnsetAt, speechOffsetAt,\n streamEndAt: clock.now(), maxFrameGapMs };\n}\n\nexport async function streamSilenceWhile(omni, rate, shouldContinue, clock, onFrame = () => {}) {\n const silence = new Int16Array(Math.max(1, Math.round((rate * FRAME_MS) / 1000)));\n while (shouldContinue()) {\n const at = clock.now();\n omni.sendAudio(silence);\n onFrame(silence, at);\n await clock.sleep(FRAME_MS);\n }\n}\n\nexport async function waitForAgentSettle(getCtx, clock, opts = {}) {\n const started = clock.now();\n const observation = opts.callerStreamEndAt != null;\n if (observation && (!Number.isFinite(opts.callerStreamEndAt)\n || opts.callerStreamEndAt < 0 || opts.callerStreamEndAt > started)) {\n throw new Error(\"Invalid caller observation boundary\");\n }\n const timeoutMs = observation ? INTERRUPTION_OBSERVATION.maximumAfterCallerStreamMs : opts.timeoutMs ?? 25000;\n const settleMs = observation ? INTERRUPTION_OBSERVATION.quietMs : opts.settleMs ?? 2000;\n const deadlineAt = (observation ? opts.callerStreamEndAt : started) + timeoutMs;\n const minimumAt = observation ? opts.callerStreamEndAt + INTERRUPTION_OBSERVATION.minimumAfterCallerStreamMs : -Infinity;\n while (true) {\n const ctx = getCtx();\n const at = clock.now();\n if (opts.isClosed?.()) return { reason: \"closed\", at };\n if (observation && at >= deadlineAt) return { reason: \"timeout\", at };\n const hasResponseAudio = !opts.requireTurnBegin || (ctx.latestTurnBeginAt != null\n && ctx.postTurnBeginFirstAudioAt != null);\n // Synthesis text is progress, not a completion event. It can arrive after\n // early/final audio, so extend the quiet floor without demanding another\n // packet after every advisory. The protocol still has no reply-end marker.\n const activityEnd = Math.max(ctx.playbackEndAt ?? -Infinity,\n ctx.latestTurnBeginAt ?? -Infinity, ctx.lastAssistantTranscriptAt ?? -Infinity,\n observation ? ctx.lastCallerTranscriptAt ?? -Infinity : -Infinity);\n // Do not demand a new post-caller turn_begin: an earlier turn can be\n // coalesced and speak later. This records a bounded quiet observation;\n // downstream response-timing checks can impose a stricter begin boundary.\n if (ctx.firstPacketAt != null && hasResponseAudio && at >= minimumAt && at >= activityEnd + settleMs) {\n return { reason: \"settled\", at };\n }\n if (opts.allowEmpty && ctx.firstPacketAt == null && at >= minimumAt && at - started >= (opts.emptyWaitMs ?? 800)) {\n return { reason: \"empty\", at };\n }\n if (at - started >= timeoutMs) return { reason: \"timeout\", at };\n await clock.sleep(observation ? Math.min(20, deadlineAt - at) : 20);\n }\n}\n\nexport async function withTimeout(promise, ms, message, clock) {\n let timer;\n try {\n return await Promise.race([\n promise,\n new Promise((_, reject) => {\n timer = clock.setTimeout(() => reject(new Error(message)), ms);\n }),\n ]);\n } finally {\n clock.clearTimeout(timer);\n }\n}\n",
|
|
6
6
|
"main.mjs": "import PyAI from \"@pyai/sdk\";\nimport WebSocket from \"ws\"; // Explicit transport makes this work on Node 20 too.\nimport { readFile, writeFile, mkdtemp } from \"node:fs/promises\";\nimport { join } from \"node:path\";\nimport { runSession } from \"./session.mjs\";\nimport { readWav, writeWav, assessAnswer } from \"./audio.mjs\";\n\nasync function main() {\n const [callerPath, interruptionPath, extra] = process.argv.slice(2);\n if (!callerPath || extra) throw new Error(\"Usage: node main.mjs caller.wav [interruption.wav]\");\n if (!process.env.PYAI_API_KEY) throw new Error(\"Inject PYAI_API_KEY through your environment first\");\n const caller = readWav(await readFile(callerPath));\n const interruption = interruptionPath ? readWav(await readFile(interruptionPath)) : undefined;\n const pyai = new PyAI({ apiKey: process.env.PYAI_API_KEY, maxRetries: 0,\n fetch: (url, options) => fetch(url, { ...options, signal: AbortSignal.timeout(30000) }),\n });\n const { report, audio, replyAudio } = await runSession({ pyai, webSocket: WebSocket, caller, interruption });\n const directory = await mkdtemp(join(process.cwd(), \"omni-run-\"));\n const save = (name, data) => writeFile(join(directory, name), data, { mode: 0o600, flag: \"wx\" });\n await save(\"session.wav\", writeWav(audio, report.output_rate ?? 24000));\n const reply = writeWav(replyAudio, report.output_rate ?? 24000);\n await save(\"reply.wav\", reply);\n // Hear the captured bytes, rather than trusting synthesis-advisory text.\n let capturedTranscript = \"\";\n if (report.reply_audio_bytes) {\n try {\n const heard = await pyai.audio.transcriptions.create({ file: new File([reply], \"reply.wav\", { type: \"audio/wav\" }), model: \"pyai-hear\" });\n capturedTranscript = heard.text;\n } catch { report.hear_error = \"Captured audio saved; Hear verification failed\"; }\n }\n const result = { ...report, captured_transcript: capturedTranscript, evidence: assessAnswer(report, capturedTranscript) };\n await save(\"report.json\", JSON.stringify(result, null, 2) + \"\\n\");\n console.log(JSON.stringify({ directory, ...result }, null, 2));\n if (result.evidence.captured_answer !== \"verified_by_hear\" || (interruption && result.evidence.interruption !== \"simulated_queue_cleared\")) process.exitCode = 1;\n}\n\nmain().catch(error => {\n const message = String(error.message).replaceAll(process.env.PYAI_API_KEY || \"\\0\", \"[REDACTED]\");\n console.error(JSON.stringify({ error: message, ...(error.omni_code ? { code: error.omni_code } : {}) })); process.exitCode = 1;\n});\n",
|
|
7
7
|
"session.mjs": "import { newAudioCapture, recordAudio, recordTurnBegin, streamPcmRealtime, INTERRUPTION_OBSERVATION } from \"./timing.mjs\";\n\nexport const realClock = {\n now: () => performance.now(),\n sleep: ms => new Promise(resolve => setTimeout(resolve, ms)),\n};\n\n// Only recognized machine codes may leave the example. Error messages, call\n// IDs and arbitrary code values can contain credentials or customer content.\nconst publicErrorCodes = new Set([\n \"invalid_configure\", \"unsupported_tool_transport\", \"tool_configuration_locked\",\n \"media_dead\", \"unauthorized\", \"insufficient_scope\", \"rate_limit_exceeded\",\n \"concurrency_limit_exceeded\", \"daily_cap_exceeded\", \"credit_exhausted\",\n]);\nconst transportErrorCodes = new Set([\"ECONNRESET\", \"ECONNREFUSED\", \"ETIMEDOUT\", \"ENOTFOUND\", \"EAI_AGAIN\"]);\nexport function omniFailure(error) {\n const code = error instanceof Error\n ? (transportErrorCodes.has(error.code) ? error.code : \"transport_or_protocol_error\")\n : (publicErrorCodes.has(error?.code) ? error.code : \"unrecognized_server_error\");\n return Object.assign(new Error(`Omni failed (${code}); no automatic retry was attempted`), { omni_code: code });\n}\n\n/** One bounded, server-side example. Its playback queue is simulated. */\nexport async function runSession({ pyai, webSocket, caller, interruption, clock = realClock, timeoutMs = 60000 }) {\n const rate = 24000, frameSize = 480, silence = new Int16Array(frameSize);\n if (!(caller instanceof Int16Array) || !caller.length || caller.length > rate * 20) throw new Error(\"Provide 1–20 seconds of caller PCM16 at 24 kHz\");\n if (interruption && (!(interruption instanceof Int16Array) || !interruption.length || interruption.length > rate * 20)) throw new Error(\"Invalid interruption PCM\");\n const started = clock.now(), all = newAudioCapture(started), reply = newAudioCapture(started);\n const report = { input_rate: rate, output_rate: null, playback_sink: \"simulated\", tool_calls: 0, tool_executions: 0, duplicate_tool_calls: 0, flushes_after_interruption: 0, cleared_queue_ms: 0, interruption_requested: !!interruption, caller_started_ms: null, interruption_started_ms: null, advisory_transcripts: [], end_reason: \"timeout\" };\n let session, configuredAt = null, closed = false, failure = null, phase = \"greeting\", offset = 0, queueEnd = started, interruptionSent = false, inputEndAt = started;\n const handled = new Set();\n const fail = message => { failure ??= new Error(message); };\n try {\n session = pyai.omni.connect({\n webSocket, rate,\n configure: {\n voice_id: \"stock_amos_en_us\", language: \"en\", greeting: \"Hello. How can I help?\",\n persona: \"When asked about office hours, call lookup_office_hours. Never guess. After the result, say one short sentence giving its opening_time.\",\n tools: [{ name: \"lookup_office_hours\", description: \"Read the synthetic example office opening time.\", side_effect: \"read\", parameters: { type: \"object\", properties: {}, required: [], additionalProperties: false } }],\n },\n onHello: frame => {\n const match = /^pcm16@(8000|24000)$/.exec(frame.audio_out ?? \"\");\n if (!match) return fail(\"Unsupported Omni output format\");\n report.output_rate = Number(match[1]);\n },\n onConfigured: () => { configuredAt ??= clock.now(); },\n onAudio: chunk => {\n if (!report.output_rate) return fail(\"Audio arrived before its output format\");\n const bytes = chunk instanceof ArrayBuffer ? new Uint8Array(chunk) : new Uint8Array(chunk.buffer, chunk.byteOffset, chunk.byteLength);\n if (bytes.byteLength % 2) return fail(\"Invalid PCM16 output\");\n const view = new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength);\n const pcm = Int16Array.from({ length: bytes.length / 2 }, (_, i) => view.getInt16(i * 2, true));\n const now = clock.now();\n recordAudio(all, pcm, now, report.output_rate);\n queueEnd = Math.max(now, queueEnd) + pcm.length / report.output_rate * 1000;\n if (report.tool_executions) recordAudio(reply, pcm, now, report.output_rate);\n },\n onEvent: frame => {\n if (frame.event === \"turn_begin\") recordTurnBegin(all, clock.now(), frame.turn);\n },\n onTranscript: frame => {\n // An assistant advisory is progress; never treat it as received speech.\n if (report.advisory_transcripts.length < 100) report.advisory_transcripts.push({ role: frame.role, text: frame.text });\n if (frame.role === \"assistant\") all.lastAssistantTranscriptAt = clock.now();\n },\n onToolCall: frame => {\n report.tool_calls++;\n if (handled.has(frame.call_id)) { report.duplicate_tool_calls++; return; }\n handled.add(frame.call_id);\n if (frame.name !== \"lookup_office_hours\") { session.toolResult(frame.call_id, { error: \"Unknown example tool\" }); return; }\n // Replace this read-only fixture with your authorized data lookup.\n report.tool_executions++;\n session.toolResult(frame.call_id, { result: { opening_time: \"9 a.m.\", source: \"synthetic example office\" } });\n },\n onBargeIn: () => {\n const now = clock.now(), queued = Math.max(0, queueEnd - now);\n if (report.interruption_started_ms != null) { report.flushes_after_interruption++; report.cleared_queue_ms += queued; }\n // A real speaker adapter must cancel scheduled and currently playing audio.\n queueEnd = now; all.playbackEndAt = now; reply.playbackEndAt = now;\n },\n onError: error => { failure ??= omniFailure(error); },\n onClose: () => { closed = true; },\n });\n while (clock.now() - started < timeoutMs) {\n if (failure) throw failure;\n if (closed) { report.end_reason = \"closed\"; break; }\n const now = clock.now();\n if (configuredAt == null || !report.output_rate) {\n if (now - started >= 10000) throw new Error(\"Omni connection/configuration timed out\");\n await clock.sleep(20); continue;\n }\n const quietAt = Math.max(queueEnd, all.lastPacketAt ?? now, all.lastAssistantTranscriptAt ?? 0, inputEndAt);\n if (phase === \"greeting\") {\n if (all.firstPacketAt != null && now - configuredAt >= 2000 && now >= quietAt + 800) {\n phase = \"caller\"; report.caller_started_ms = now - started;\n } else if (now - configuredAt > 15000) throw new Error(\"Greeting did not drain within 15 seconds\");\n }\n if (phase === \"reply\" && interruption && !interruptionSent && reply.samples && queueEnd - now >= 200) {\n phase = \"interruption\"; offset = 0; interruptionSent = true; report.interruption_started_ms = now - started;\n }\n const observationEnd = inputEndAt + (interruptionSent ? INTERRUPTION_OBSERVATION.minimumAfterCallerStreamMs : 0);\n if (phase === \"reply\" && reply.samples && now >= quietAt + 2000 && now >= observationEnd && (!interruption || interruptionSent)) {\n report.end_reason = \"quiet_window\"; break;\n }\n // Exactly one producer: each 20 ms slot contains caller PCM OR silence.\n let pcm = silence;\n if (phase === \"caller\" || phase === \"interruption\") {\n const source = phase === \"caller\" ? caller : interruption;\n pcm = source.subarray(offset, offset + frameSize); offset += pcm.length;\n if (offset >= source.length) {\n inputEndAt = now + pcm.length / rate * 1000;\n report[phase === \"caller\" ? \"caller_completed_ms\" : \"interruption_completed_ms\"] = inputEndAt - started;\n phase = \"reply\"; offset = 0;\n }\n }\n await streamPcmRealtime(session, pcm, rate, clock);\n }\n report.audio_bytes = all.samples * 2;\n report.reply_audio_bytes = reply.samples * 2;\n report.elapsed_ms = Math.round(clock.now() - started);\n return { report, audio: all.pcm, replyAudio: reply.pcm };\n } finally { session?.close(); }\n}\n",
|
|
@@ -88,7 +88,7 @@ export class PyaiDialerElement extends HTMLElement {
|
|
|
88
88
|
const complete = this.root.querySelector<HTMLButtonElement>("[data-action=complete-transfer]")!;
|
|
89
89
|
complete.hidden = transfer?.mode !== "warm" || transfer.status !== "ready"; complete.disabled = Boolean(snapshot.transferPending);
|
|
90
90
|
const cancel = this.root.querySelector<HTMLButtonElement>("[data-action=cancel-transfer]")!;
|
|
91
|
-
cancel.hidden = transfer
|
|
91
|
+
cancel.hidden = !transfer || !pending; cancel.disabled = Boolean(snapshot.transferPending || ["completing", "canceling"].includes(transfer?.status ?? ""));
|
|
92
92
|
this.root.querySelector<HTMLButtonElement>("[data-action=recover-transfer]")!.hidden = !snapshot.transferPending;
|
|
93
93
|
}
|
|
94
94
|
private async run(action: () => Promise<void>) {
|
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import type { TelephonyBrowserSession, TelephonyBrowserSessionCreate, TelephonyCall } from "../index.ts";
|
|
1
|
+
import type { TelephonyBrowserSession, TelephonyBrowserSessionCreate, TelephonyCall, TelephonyCallControl } from "../index.ts";
|
|
2
2
|
|
|
3
3
|
/** Implement on your server using PyAI's telephony client. The browser receives
|
|
4
4
|
* a single-call capability, never an organization key or carrier credential. */
|
|
@@ -7,6 +7,7 @@ export interface DialerBroker {
|
|
|
7
7
|
get(callId: string): Promise<TelephonyCall>;
|
|
8
8
|
hangup(callId: string): Promise<TelephonyCall>;
|
|
9
9
|
setHold?(callId: string, held: boolean, key: string): Promise<TelephonyCall>;
|
|
10
|
+
sendDtmf?(callId: string, digits: string, key: string): Promise<TelephonyCallControl>;
|
|
10
11
|
transfers?: {
|
|
11
12
|
create(callId: string, input: { to: string; mode: "warm" | "cold" }, key: string): Promise<TelephonyCall>;
|
|
12
13
|
complete(callId: string, transferId: string): Promise<TelephonyCall>;
|
|
@@ -45,7 +46,7 @@ export interface DialerSnapshot {
|
|
|
45
46
|
const terminal = (call: TelephonyCall) => ["completed", "failed", "canceled"].includes(call.status);
|
|
46
47
|
/** A durable handoff keeps its original room after the browser goes away. */
|
|
47
48
|
export const dialerCallHandedOff = (call: TelephonyCall) => call.transfer_enabled === true && Boolean(call.transfers?.some(transfer =>
|
|
48
|
-
transfer.status === "completed" || transfer.mode === "cold" &&
|
|
49
|
+
transfer.status === "completed" || transfer.mode === "cold" && transfer.status === "completing"));
|
|
49
50
|
|
|
50
51
|
/** One immutable workspace/user scope per instance. Setup and stop retain the
|
|
51
52
|
* original intent through lost acknowledgments, reloads and audio failures. */
|
|
@@ -258,7 +259,8 @@ export class ManagedDialer {
|
|
|
258
259
|
if (!this.intent?.transfer?.id || !this.broker.transfers) throw new Error("Recover the transfer identity first.");
|
|
259
260
|
const transfer = this.snapshot.call?.transfers?.find(item => item.id === this.intent!.transfer!.id);
|
|
260
261
|
if (command === "complete" && transfer?.status !== "ready") throw new Error("Wait until the transfer target has joined.");
|
|
261
|
-
if (this.intent.transfer.mode !== "warm") throw new Error("Cold transfers complete automatically.");
|
|
262
|
+
if (this.intent.transfer.mode !== "warm" && command === "complete") throw new Error("Cold transfers complete automatically.");
|
|
263
|
+
if (command === "cancel" && ["completing", "completed"].includes(transfer?.status ?? "")) throw new Error("The handoff has started. Check its status instead of canceling it.");
|
|
262
264
|
if (this.intent.transfer.command && this.intent.transfer.command !== command) throw new Error("Recover the saved transfer command first.");
|
|
263
265
|
const next = structuredClone(this.intent); next.transfer!.command = command;
|
|
264
266
|
await this.journal.write(next); this.intent = next;
|
|
@@ -2,7 +2,6 @@
|
|
|
2
2
|
import { createHash } from "node:crypto";
|
|
3
3
|
import type { PyAI, TelephonyBrowserSessionCreate } from "../index.ts";
|
|
4
4
|
import type { DialerBroker } from "./managedDialer.ts";
|
|
5
|
-
import { dialerCallHandedOff } from "./managedDialer.ts";
|
|
6
5
|
|
|
7
6
|
export function createServerDialerBroker(sdk: PyAI, authenticatedUserId: string): DialerBroker {
|
|
8
7
|
if (!authenticatedUserId.trim()) throw new Error("An authenticated user is required.");
|
|
@@ -20,9 +19,9 @@ export function createServerDialerBroker(sdk: PyAI, authenticatedUserId: string)
|
|
|
20
19
|
},
|
|
21
20
|
get: ownedCall,
|
|
22
21
|
async hangup(id: string) {
|
|
23
|
-
|
|
24
|
-
|
|
25
|
-
return sdk.telephony.calls.hangup(id);
|
|
22
|
+
await ownedCall(id);
|
|
23
|
+
// The provider rechecks handoff ownership in the stop CAS, after this read.
|
|
24
|
+
return sdk.telephony.calls.hangup(id,{user_id:authenticatedUserId});
|
|
26
25
|
},
|
|
27
26
|
async setHold(id, held, key) {
|
|
28
27
|
await ownedCall(id);
|
|
@@ -30,6 +29,12 @@ export function createServerDialerBroker(sdk: PyAI, authenticatedUserId: string)
|
|
|
30
29
|
const scopedKey = "dialer-hold:" + createHash("sha256").update(authenticatedUserId).update("\0").update(id).update("\0").update(key).digest("hex");
|
|
31
30
|
return sdk.telephony.calls.setHold(id, { held, user_id: authenticatedUserId }, { idempotencyKey: scopedKey });
|
|
32
31
|
},
|
|
32
|
+
async sendDtmf(id, digits, key) {
|
|
33
|
+
await ownedCall(id);
|
|
34
|
+
if (!key || key.length > 255 || /[\r\n]/.test(key)) throw new Error("A valid keypad retry key is required.");
|
|
35
|
+
const scopedKey = "dialer-keypad:" + createHash("sha256").update(authenticatedUserId).update("\0").update(id).update("\0").update(key).digest("hex");
|
|
36
|
+
return sdk.telephony.calls.sendDtmf(id, { digits, user_id: authenticatedUserId }, { idempotencyKey: scopedKey });
|
|
37
|
+
},
|
|
33
38
|
transfers: {
|
|
34
39
|
async create(id, input, key) {
|
|
35
40
|
await ownedCall(id);
|
package/src/index.ts
CHANGED
|
@@ -1504,6 +1504,7 @@ export interface TelephonyCall {
|
|
|
1504
1504
|
held?: boolean;
|
|
1505
1505
|
hold_pending?: boolean;
|
|
1506
1506
|
transfers?: TelephonyTransfer[];
|
|
1507
|
+
forwarding?: { to: string | null; status: "waiting" | "claimed" | "unknown" | "acknowledged" | "canceled"; menu?: { kind: "keypad"; status: "pending" | "selected" | "rejected"; choice_id: string | null } };
|
|
1507
1508
|
user_id?: string;
|
|
1508
1509
|
client_reference: string | null;
|
|
1509
1510
|
stop_requested: boolean;
|
|
@@ -1546,6 +1547,14 @@ export interface TelephonyCallCapabilities {
|
|
|
1546
1547
|
revalidation_required: true;
|
|
1547
1548
|
}
|
|
1548
1549
|
|
|
1550
|
+
export interface TelephonyCallControl {
|
|
1551
|
+
object: "telephony.call_control";
|
|
1552
|
+
id: string;
|
|
1553
|
+
call_id: string;
|
|
1554
|
+
action: "send_dtmf";
|
|
1555
|
+
status: "acknowledged" | "unknown" | "rejected";
|
|
1556
|
+
}
|
|
1557
|
+
|
|
1549
1558
|
export interface TelephonyCallCreate {
|
|
1550
1559
|
from_number_id: string;
|
|
1551
1560
|
to: string;
|
|
@@ -1593,6 +1602,12 @@ export interface TelephonyAvailableNumber {
|
|
|
1593
1602
|
region?: string | null;
|
|
1594
1603
|
capabilities?: TelephonyCapabilities;
|
|
1595
1604
|
monthly_cost_cents?: number;
|
|
1605
|
+
/** Bind purchase to the exact account/market inventory result. */
|
|
1606
|
+
inventory_reference?: string;
|
|
1607
|
+
}
|
|
1608
|
+
|
|
1609
|
+
export interface TelephonyNumberInventory extends ListResponse<TelephonyAvailableNumber> {
|
|
1610
|
+
provisioning?: { mode: "automated" | "assisted"; capability: string; message: string };
|
|
1596
1611
|
}
|
|
1597
1612
|
|
|
1598
1613
|
export interface TelephonyNumber {
|
|
@@ -1617,7 +1632,7 @@ export interface TelephonyNumberOrder {
|
|
|
1617
1632
|
id: string;
|
|
1618
1633
|
status: "pending" | "unknown" | "completed" | "failed" | "released";
|
|
1619
1634
|
phone_number: string;
|
|
1620
|
-
country: "AU";
|
|
1635
|
+
country: "AU" | "US" | "CA";
|
|
1621
1636
|
project_id: string;
|
|
1622
1637
|
retry_after_seconds: number;
|
|
1623
1638
|
message: string;
|
|
@@ -1626,6 +1641,8 @@ export interface TelephonyNumberOrder {
|
|
|
1626
1641
|
}
|
|
1627
1642
|
export interface TelephonyNumberPurchase {
|
|
1628
1643
|
phone_number: string;
|
|
1644
|
+
country?: "US" | "CA" | "IN" | "AU";
|
|
1645
|
+
inventory_reference?: string;
|
|
1629
1646
|
agent_id?: string | null;
|
|
1630
1647
|
provisioning_mode?: "automated" | "adopt_preowned";
|
|
1631
1648
|
project_id?: string;
|
|
@@ -1633,6 +1650,51 @@ export interface TelephonyNumberPurchase {
|
|
|
1633
1650
|
max_monthly_cost_cents?: number;
|
|
1634
1651
|
}
|
|
1635
1652
|
|
|
1653
|
+
export interface TelephonyInboundBinding {
|
|
1654
|
+
object: "telephony.inbound_binding";
|
|
1655
|
+
number_id: string;
|
|
1656
|
+
version: number;
|
|
1657
|
+
enabled: boolean;
|
|
1658
|
+
project_id?: string;
|
|
1659
|
+
agent_id?: string;
|
|
1660
|
+
agent_release_id?: string;
|
|
1661
|
+
agent_profile_sha256?: string;
|
|
1662
|
+
admission_mode?: "client" | "platform";
|
|
1663
|
+
forward?: TelephonyInboundForward | null;
|
|
1664
|
+
carrier_operation?: { id: string; phase: "pending" | "unknown" | "active" | "rollback_pending" | "rolled_back"; action: "activate" | "rollback" };
|
|
1665
|
+
}
|
|
1666
|
+
|
|
1667
|
+
export interface TelephonyInboundMenu {
|
|
1668
|
+
/** Saved single-digit spoken menu before forwarding. */
|
|
1669
|
+
kind: "keypad";
|
|
1670
|
+
prompt: string;
|
|
1671
|
+
choices: Array<{ id: string; key: string; to: string }>;
|
|
1672
|
+
timeout_seconds: number;
|
|
1673
|
+
/** null hangs up; otherwise the ID of one saved choice. */
|
|
1674
|
+
timeout_choice_id: string | null;
|
|
1675
|
+
}
|
|
1676
|
+
export type TelephonyInboundForward = {
|
|
1677
|
+
max_duration_seconds: number;
|
|
1678
|
+
recording: boolean;
|
|
1679
|
+
} & ({ to: string; menu?: never } | { to?: never; menu: TelephonyInboundMenu });
|
|
1680
|
+
|
|
1681
|
+
export type TelephonyInboundBindingUpdate = {
|
|
1682
|
+
expected_version: number;
|
|
1683
|
+
activate_carrier_route?: boolean;
|
|
1684
|
+
rollback_carrier_route?: boolean;
|
|
1685
|
+
/** Persist the original operation ID; recover it instead of inventing a new one. */
|
|
1686
|
+
carrier_operation_id?: string;
|
|
1687
|
+
} & ({ enabled: false } | {
|
|
1688
|
+
enabled: true;
|
|
1689
|
+
project_id: string;
|
|
1690
|
+
agent_id: string;
|
|
1691
|
+
agent_release_id: string;
|
|
1692
|
+
agent_profile_sha256: string;
|
|
1693
|
+
admission_mode?: "client" | "platform";
|
|
1694
|
+
agent_knowledge_receipt?: Record<string, unknown>;
|
|
1695
|
+
forward?: TelephonyInboundForward | null;
|
|
1696
|
+
});
|
|
1697
|
+
|
|
1636
1698
|
/* ------------------------------------------------------------------------- *
|
|
1637
1699
|
* Trace (compliance)
|
|
1638
1700
|
* ------------------------------------------------------------------------- */
|
|
@@ -2432,7 +2494,7 @@ export class PyAI {
|
|
|
2432
2494
|
return this.getJson(`/v1/telephony/calls${query.size ? `?${query}` : ""}`);
|
|
2433
2495
|
},
|
|
2434
2496
|
/** Durable stop intent; use get() until a terminal status is confirmed. */
|
|
2435
|
-
hangup: (id: string): Promise<TelephonyCall> => this.postJson(`/v1/telephony/calls/${encodeURIComponent(id)}/hangup`, {}),
|
|
2497
|
+
hangup: (id: string, input?: {user_id:string}): Promise<TelephonyCall> => this.postJson(`/v1/telephony/calls/${encodeURIComponent(id)}/hangup`, input ?? {}),
|
|
2436
2498
|
decideAdmission: (id: string, input: { decision: "accept" | "reject"; client_reference: string; max_duration_seconds: number }): Promise<TelephonyCall> =>
|
|
2437
2499
|
this.postJson(`/v1/telephony/calls/${encodeURIComponent(id)}/admission`, input),
|
|
2438
2500
|
recording: (id: string, options: { range?: string } = {}): Promise<Response> =>
|
|
@@ -2441,6 +2503,12 @@ export class PyAI {
|
|
|
2441
2503
|
/** The same durable key recovers an uncertain hold command without repeating it. */
|
|
2442
2504
|
setHold: (id: string, input: { user_id: string; held: boolean }, options: { idempotencyKey: string }): Promise<TelephonyCall> =>
|
|
2443
2505
|
this.postJson(`/v1/telephony/calls/${encodeURIComponent(id)}/hold`, input, telephonyIdempotencyHeader(options.idempotencyKey)),
|
|
2506
|
+
/** Server-side keypad command for an owned switch call. Reuse the same key
|
|
2507
|
+
* after an unknown result; never store keypad digits in a browser journal. */
|
|
2508
|
+
sendDtmf: (id: string, input: { digits: string; user_id: string }, options: { idempotencyKey: string }): Promise<TelephonyCallControl> => {
|
|
2509
|
+
if (!/^[0-9A-D*#]{1,32}$/.test(input.digits)) throw new Error("Use 1–32 keypad digits (0–9, *, #, A–D).");
|
|
2510
|
+
return this.postJson(`/v1/telephony/calls/${encodeURIComponent(id)}/dtmf`, input, telephonyIdempotencyHeader(options.idempotencyKey));
|
|
2511
|
+
},
|
|
2444
2512
|
transfers: {
|
|
2445
2513
|
/** Persist the key and target before requesting a consultation. */
|
|
2446
2514
|
create: (id: string, input: { to: string; mode: "warm" | "cold"; user_id: string }, options: { idempotencyKey: string }): Promise<TelephonyCall> =>
|
|
@@ -2460,7 +2528,7 @@ export class PyAI {
|
|
|
2460
2528
|
/** Search managed inventory; country availability and ordering may be gated. */
|
|
2461
2529
|
available: (
|
|
2462
2530
|
params: { country?: "US" | "CA" | "IN" | "AU"; areaCode?: string; contains?: string; limit?: number } = {},
|
|
2463
|
-
): Promise<
|
|
2531
|
+
): Promise<TelephonyNumberInventory> => {
|
|
2464
2532
|
const q = new URLSearchParams();
|
|
2465
2533
|
if (params.country) q.set("country", params.country);
|
|
2466
2534
|
if (params.areaCode) q.set("area_code", params.areaCode);
|
|
@@ -2480,6 +2548,12 @@ export class PyAI {
|
|
|
2480
2548
|
/** Route a number to an agent (`agentId: null` to unassign). */
|
|
2481
2549
|
assign: (id: string, agentId: string | null): Promise<TelephonyNumber> =>
|
|
2482
2550
|
this.postJson(`/v1/telephony/numbers/${encodeURIComponent(id)}/assign`, { agent_id: agentId }),
|
|
2551
|
+
getInboundBinding: (id: string): Promise<TelephonyInboundBinding> =>
|
|
2552
|
+
this.getJson(`/v1/telephony/numbers/${encodeURIComponent(id)}/inbound-binding`),
|
|
2553
|
+
/** Save the profile and stable operation first; after uncertainty read the
|
|
2554
|
+
* binding before explicitly recovering that same operation. */
|
|
2555
|
+
setInboundBinding: (id: string, input: TelephonyInboundBindingUpdate): Promise<TelephonyInboundBinding> =>
|
|
2556
|
+
this.putJson(`/v1/telephony/numbers/${encodeURIComponent(id)}/inbound-binding`, input),
|
|
2483
2557
|
/** Release a number back to the carrier (idempotent). */
|
|
2484
2558
|
release: async (id: string): Promise<TelephonyNumber> => {
|
|
2485
2559
|
const res = await this.deleteReq(`/v1/telephony/numbers/${encodeURIComponent(id)}`);
|