@pyai/sdk 0.17.0 → 0.18.1

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/src/index.ts CHANGED
@@ -126,10 +126,13 @@ export type SpeakModel = "pyai-speak" | "tts-1" | "tts-1-hd";
126
126
 
127
127
  export interface SpeechParams {
128
128
  input: string;
129
+ /** Reuse one unique call ID across utterances for enabled call-stable voice recovery. */
130
+ speech_session_id?: string;
129
131
  voice?: string;
130
132
  model?: SpeakModel;
131
133
  /**
132
- * Delivery mode. The API defaults to true and every catalog voice accepts
134
+ * Delivery mode. `speech` defaults to false; `speechStream` always uses true.
135
+ * The API itself defaults to true and every catalog voice accepts
133
136
  * both values. Set false when you want one complete buffered body with a
134
137
  * `Content-Length`. A voice whose serving fleet has no streaming lane is
135
138
  * rendered buffered either way and says so with `x-pyai-stream: buffered`.
@@ -183,6 +186,10 @@ export type CreateJobParams = CreateJobOptions & (
183
186
  );
184
187
 
185
188
  export interface CreateJobOptions {
189
+ /** Spoken-language hint for every audio chunk; omit for automatic detection. */
190
+ transcription_language?: string;
191
+ /** Original script (default), or Roman letters preserving the spoken language; no translation. */
192
+ output_script?: "original" | "latin";
186
193
  /** Requires Trace entitlement; incompatible with channel/diarize. */
187
194
  trace?: boolean;
188
195
  rule_pack?: Record<string, unknown>;
@@ -500,7 +507,7 @@ export interface HearStreamOptions {
500
507
  /** Streaming STT model. Server default `pyai-hear`. */
501
508
  model?: string;
502
509
  /** Omit or use auto for automatic detection; an explicit code pins recognition. */
503
- language?: "auto" | "en" | "es" | "fr" | "de" | "hi" | "it" | "pt" | "nl" | "zh" | "yue" | "ar" | "id" | "ko" | "th" | "vi" | "ja" | "tr" | "ms" | "sv" | "da" | "fi" | "pl" | "cs" | "fil" | "fa" | "el" | "hu" | "mk" | "ro";
510
+ language?: "auto" | "en" | "es" | "fr" | "de" | "hi" | "it" | "pt" | "nl" | "zh" | "yue" | "ar" | "id" | "ko" | "th" | "vi" | "ja" | "tr" | "ms" | "sv" | "da" | "fi" | "pl" | "cs" | "fil" | "fa" | "el" | "hu" | "mk" | "ro" | "ru";
504
511
  /** Input PCM sample rate in Hz. Default 16000 server-side. */
505
512
  sampleRate?: number;
506
513
  /** Audio frame encoding. Default "pcm16". */
@@ -1650,6 +1657,21 @@ export interface TelephonyNumberPurchase {
1650
1657
  max_monthly_cost_cents?: number;
1651
1658
  }
1652
1659
 
1660
+ export interface TelephonyDesktopEndpoint {
1661
+ object:'telephony.desktop_endpoint';id:string;project_id:string;user_id:string;actor_type:'application';label:string;
1662
+ registered:boolean;available:boolean;number_id:string;status:'active'|'revoked';
1663
+ }
1664
+ export interface TelephonyDesktopEnrollment {object:'telephony.desktop_enrollment';version:1;endpoint_id:string;enrollment_token:string;expires_at:number;redeem_path:string;}
1665
+ export interface TelephonyDesktopCredentials {endpoint_id:string;org_id:string;project_id:string;user_id:string;label:string;refresh_token:string;refresh_expires_at:number;access_token:string;expires_at:number;presence_token:string;heartbeat_interval_seconds:15;}
1666
+ export interface TelephonyInboundRingGroup {
1667
+ mode:'simultaneous'|'desktop_first';desktop_endpoint_ids:[string];external_to:string;desktop_timeout_seconds:number;
1668
+ ring_timeout_seconds:number;on_no_answer:'hangup';max_duration_seconds:number;recording:boolean;
1669
+ }
1670
+ export interface TelephonyInboundCapabilities {
1671
+ object:'telephony.inbound_capabilities';number_id:string;binding_version:number;
1672
+ ring_group:{supported:boolean;modes:Array<'simultaneous'|'desktop_first'>;max_desktop_endpoints:1;on_no_answer:['hangup'];desktop_registration_required:true};
1673
+ }
1674
+
1653
1675
  export interface TelephonyInboundBinding {
1654
1676
  object: "telephony.inbound_binding";
1655
1677
  number_id: string;
@@ -1661,6 +1683,7 @@ export interface TelephonyInboundBinding {
1661
1683
  agent_profile_sha256?: string;
1662
1684
  admission_mode?: "client" | "platform";
1663
1685
  forward?: TelephonyInboundForward | null;
1686
+ ring_group?:TelephonyInboundRingGroup|null;
1664
1687
  carrier_operation?: { id: string; phase: "pending" | "unknown" | "active" | "rollback_pending" | "rolled_back"; action: "activate" | "rollback" };
1665
1688
  }
1666
1689
 
@@ -1693,6 +1716,7 @@ export type TelephonyInboundBindingUpdate = {
1693
1716
  admission_mode?: "client" | "platform";
1694
1717
  agent_knowledge_receipt?: Record<string, unknown>;
1695
1718
  forward?: TelephonyInboundForward | null;
1719
+ ring_group?:TelephonyInboundRingGroup|null;
1696
1720
  });
1697
1721
 
1698
1722
  /* ------------------------------------------------------------------------- *
@@ -2320,13 +2344,13 @@ export class PyAI {
2320
2344
  // --- audio --------------------------------------------------------------
2321
2345
 
2322
2346
  audio = {
2323
- /** Text-to-speech. Returns the raw audio bytes (default WAV). */
2347
+ /** Text-to-speech. Returns complete audio bytes (default WAV, stream:false). */
2324
2348
  speech: async (params: SpeechParams): Promise<ArrayBuffer> => {
2325
2349
  assertActiveSpeechParams(params);
2326
2350
  const res = await this.request("/v1/audio/speech", {
2327
2351
  method: "POST",
2328
2352
  headers: this.authHeaders({ "Content-Type": "application/json" }),
2329
- body: JSON.stringify({ model: "pyai-speak", ...params }),
2353
+ body: JSON.stringify({ model: "pyai-speak", ...params, stream: params.stream ?? false }),
2330
2354
  });
2331
2355
  await validateSpeechResponse(res, params);
2332
2356
  return res.arrayBuffer();
@@ -2417,11 +2441,15 @@ export class PyAI {
2417
2441
  },
2418
2442
  /** Upload a recording with a durable retry key. Requires the current
2419
2443
  * keyed multipart job contract; never change the key after an unknown ACK. */
2420
- upload: async (params: { file: Blob; filename: string; call_id?: string; diarize?: boolean; output_formats?: Array<"json" | "srt" | "vtt"> }, opts: { idempotencyKey: string }): Promise<TranscriptionJob> => {
2444
+ upload: async (params: { file: Blob; filename: string; call_id?: string; diarize?: boolean; channel?: boolean; transcription_language?: string; output_script?: "original" | "latin"; webhook_url?: string; output_formats?: Array<"json" | "srt" | "vtt"> }, opts: { idempotencyKey: string }): Promise<TranscriptionJob> => {
2421
2445
  if (!opts.idempotencyKey || opts.idempotencyKey.length > 255 || /[\r\n]/.test(opts.idempotencyKey)) throw new Error("A valid recording retry key is required.");
2422
2446
  const form = new FormData();
2423
2447
  if (params.call_id) form.set("call_id", params.call_id);
2424
2448
  if (params.diarize !== undefined) form.set("diarize", String(params.diarize));
2449
+ if (params.channel !== undefined) form.set("channel", String(params.channel));
2450
+ if (params.transcription_language) form.set("transcription_language", params.transcription_language);
2451
+ if (params.output_script) form.set("output_script", params.output_script);
2452
+ if (params.webhook_url) form.set("webhook_url", params.webhook_url);
2425
2453
  if (params.output_formats) form.set("output_formats", params.output_formats.join(","));
2426
2454
  form.set("audio", params.file, params.filename);
2427
2455
  const response = await this.request("/v1/transcription/jobs", { method: "POST", headers: this.authHeaders({ "Idempotency-Key": opts.idempotencyKey }), body: form });
@@ -2482,6 +2510,11 @@ export class PyAI {
2482
2510
 
2483
2511
  telephony = {
2484
2512
  /** Server-side methods. Keep the PyAI key in your backend. */
2513
+ desktopEndpoints: {
2514
+ list:(input:{number_id?:string}={}):Promise<ListResponse<TelephonyDesktopEndpoint>>=>this.getJson('/v1/telephony/desktop-endpoints'+(input.number_id?'?'+new URLSearchParams({number_id:input.number_id}):'')),
2515
+ create:(input:{number_id:string;user_id:string;client_reference:string;label:string}):Promise<TelephonyDesktopEndpoint>=>this.postJson('/v1/telephony/desktop-endpoints',input),
2516
+ enroll:(id:string,input:{user_id:string},options:{idempotencyKey:string}):Promise<TelephonyDesktopEnrollment>=>this.postJson(`/v1/telephony/desktop-endpoints/${encodeURIComponent(id)}/enrollment`,input,telephonyIdempotencyHeader(options.idempotencyKey)),
2517
+ },
2485
2518
  calls: {
2486
2519
  create: (input: TelephonyCallCreate, options: { idempotencyKey: string }): Promise<TelephonyCall> =>
2487
2520
  this.postJson("/v1/telephony/calls", input, telephonyIdempotencyHeader(options.idempotencyKey)),
@@ -2548,6 +2581,7 @@ export class PyAI {
2548
2581
  /** Route a number to an agent (`agentId: null` to unassign). */
2549
2582
  assign: (id: string, agentId: string | null): Promise<TelephonyNumber> =>
2550
2583
  this.postJson(`/v1/telephony/numbers/${encodeURIComponent(id)}/assign`, { agent_id: agentId }),
2584
+ getInboundCapabilities:(id:string):Promise<TelephonyInboundCapabilities>=>this.getJson(`/v1/telephony/numbers/${encodeURIComponent(id)}/inbound-capabilities`),
2551
2585
  getInboundBinding: (id: string): Promise<TelephonyInboundBinding> =>
2552
2586
  this.getJson(`/v1/telephony/numbers/${encodeURIComponent(id)}/inbound-binding`),
2553
2587
  /** Save the profile and stable operation first; after uncertainty read the