@bojackduy/opencode-voice 0.11.1 → 0.11.2

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 CHANGED
@@ -18,13 +18,13 @@ One package runs on both OpenCode 1.x and 2.x. The plugin detects the host
18
18
  by capability (never by version string) and adapts; the v1 path is unchanged.
19
19
 
20
20
  > [!IMPORTANT]
21
- > **Submit-only mode on v2.** OpenCode v2.0.26 exposes no plugin API for
22
- > appending text to the composer for editing, so the v1 STT loop of
23
- > transcribe → append → edit → submit is unimplementable there. On v2 every
24
- > STT turn **transcribes and submits immediately** via `session.prompt` —
25
- > both `/stt-record` and `/stt-submit` submit — and the turn toast says
26
- > submitted, never "added to prompt". A warning toast on first load says the
27
- > same. There is deliberately no edit step; nothing is faked.
21
+ > **Dictation inserts; it does not send.** On both v1 and v2 the transcript
22
+ > is typed at the cursor of whichever editable field currently has focus —
23
+ > the session composer, a dialog/search input, or another plugin's text
24
+ > field — leaving the text editable. Nothing is submitted unless you asked
25
+ > for it (`/stt-submit`, `sttAutoSubmit`, or conversation mode). If no
26
+ > editable field is focused, v2 keeps the transcript in a recovery dialog
27
+ > instead of sending it as a prompt.
28
28
 
29
29
  ### v1 (OpenCode 1.x)
30
30
 
@@ -84,21 +84,25 @@ capability detection already picks the right path per host.
84
84
  every one reliably; the `Ctrl+P` palette shows them for matching queries
85
85
  but its search occasionally reports no results for queries that should
86
86
  match — when in doubt, use `/slash`.
87
- - **STT (batch):** submits immediately, no edit step (see above). Recording
88
- from home creates a session only when final text is ready, opens it, and
89
- submits once. The selected model is carried into the new session; existing
90
- sessions retain their agent/model. On home, v2's public plugin API exposes
91
- the selected model but no selected-agent getter, so session creation uses
92
- the host's default agent. Create/submit failures show the full transcript in
93
- a recovery dialog. Cancelling before submission drops the captured text.
87
+ - **STT (batch):** the transcript is typed at the cursor of the focused
88
+ editable field and nothing is sent unless submission is requested.
89
+ On home the composer is focused, so dictation works before a session
90
+ exists; an explicit submit creates the session, carries the selected
91
+ model into it, and sends once. Existing sessions retain their
92
+ agent/model. On home, v2's public plugin API exposes the selected model
93
+ but no selected-agent getter, so session creation uses the host's
94
+ default agent. Create/submit failures show the full transcript in a
95
+ recovery dialog. With no editable field focused, the transcript is kept
96
+ in that dialog instead of being sent as a prompt.
94
97
  - **STT (streaming dictation):** enable with `"sttMode": "streaming"` in plugin
95
98
  options. Local whisper streaming displays a bounded live transcript above
96
99
  the prompt footer on home and session screens. Stable words are plain;
97
100
  the tentative tail is bracketed and may change. Press the record key again
98
- to finalize and submit once, or cancel to clear the preview without sending.
99
- No composer edit step is promised on v2. Batch remains the default and
100
- produces text after recording stops; the hosted `sttEndpoint` option is for
101
- batch transcription, not this local streaming preview.
101
+ to finalize: the full text is inserted once at the focused cursor (and
102
+ submitted when `sttAutoSubmit` is set), or cancel to clear the preview
103
+ without sending. Batch remains the default and produces text after
104
+ recording stops; the hosted `sttEndpoint` option is for batch
105
+ transcription, not this local streaming preview.
102
106
  - **TTS / conversation / live notes / cancel:** same semantics as v1 (auto
103
107
  TTS on idle/permission/question gates, conversation barge-in, live-notes
104
108
  background saves, one-key cancel).
@@ -747,7 +751,7 @@ Idle invocations toast `Nothing to cancel`. Keybind override:
747
751
 
748
752
  Live local dictation for short prompts: the mic stays on, partial
749
753
  transcriptions stream in as you speak, and one press of `/stt-record`
750
- finalizes the text into the prompt box on v1, or submits it once on v2.
754
+ finalizes the text at the focused cursor on both hosts.
751
755
  Set `"sttMode": "streaming"` in the
752
756
  plugin options (default `"batch"` keeps the record-then-transcribe flow
753
757
  exactly as before).
@@ -764,18 +768,18 @@ there is never a silent gap — `Loading speech model…` during cold start,
764
768
  (plus a `(listening…)` note when inference runs slow), and `Finalizing…`
765
769
  between the second keypress and finalization. On v2, a bounded transcript
766
770
  above the prompt footer updates during recording: stable words are plain,
767
- tentative words are bracketed. Finalization submits directly to the active
768
- session (or creates one from home); cancel clears the preview without sending.
769
- On v1, the OpenCode renderer only
771
+ tentative words are bracketed. Finalization inserts the full text once at
772
+ the focused cursor; cancel clears the preview without sending.
773
+ The OpenCode renderer only
770
774
  exposes `insertText`/`submit`, with no live range-replacement API, so
771
775
  partials render as preview toasts (`🎙 ...`) alongside (not instead of)
772
776
  the sticky status, and the full text is inserted EXACTLY ONCE at
773
- finalize. If no editable field is focused, the transcript is kept via
774
- toast instead of being redirected into the chat. `/stt-submit` submits
775
- only through the dictated field's own `submit()` — a field that cannot
776
- submit keeps its text un-submitted rather than falling through to the
777
- primary chat. `/stt-stop` cancels and removes only plugin-dictated
778
- text, never your typed prefix/suffix.
777
+ finalize. If no editable field is focused, the transcript is kept visibly
778
+ (v2 recovery dialog, v1 toast) instead of being redirected into the chat.
779
+ `/stt-submit` submits only through the dictated field's own `submit()` — a
780
+ field that cannot submit keeps its text un-submitted rather than falling
781
+ through to the primary chat. `/stt-stop` cancels and removes only
782
+ plugin-dictated text, never your typed prefix/suffix.
779
783
 
780
784
  Warm server: the whisper model stays loaded across dictations (finalize
781
785
  and cancel keep the lease); it reloads only on model/language change,
@@ -796,9 +800,8 @@ Options:
796
800
  - `sttMode` _(optional, default `"batch"`)_ - `"streaming"` selects this mode
797
801
  - `sttStreamWindowMs` _(optional, default `10000`)_ - rolling audio window
798
802
  - `sttStreamStepMs` _(optional, default `1000`)_ - live transcription cadence
799
- - `sttAutoSubmit` applies after streaming finalize on v1 (submit via the
800
- dictated field only, as above). v2 always submits final text because there
801
- is no composer append/edit API.
803
+ - `sttAutoSubmit` applies after streaming finalize on both hosts (submit via
804
+ the dictated field only, as above).
802
805
 
803
806
  Measured latencies (Apple M3 Pro, 36 GB, whisper.cpp 1.9.4 with Metal,
804
807
  `ggml-large-v3-turbo-q5_0.bin`, scripted speech through the real
@@ -839,10 +842,11 @@ real streaming controller in `test/v2-preview.test.js`.
839
842
  3. LLM normalizes the transcription: fixes punctuation, removes filler words,
840
843
  corrects software engineering homophones ("Jason" to "JSON", "bullion" to
841
844
  "boolean", etc.)
842
- 4. Cleaned text is appended to the OpenCode prompt, or submitted immediately
843
- when `/stt-submit` is used. If normalization fails (e.g. LLM endpoint
844
- unreachable), the raw transcription is used as a fallback so you never lose
845
- your input
845
+ 4. Cleaned text is inserted at the cursor of the focused editable field
846
+ (composer, search, or plugin input), or submitted immediately when
847
+ `/stt-submit` (or `sttAutoSubmit`) is used. If normalization fails (e.g.
848
+ LLM endpoint unreachable), the raw transcription is used as a fallback
849
+ so you never lose your input
846
850
 
847
851
  ### TTS pipeline
848
852
 
package/index.js CHANGED
@@ -1,9 +1,9 @@
1
1
  // opencode-voice: Speech-to-text and text-to-speech for OpenCode.
2
2
  //
3
3
  // STT: Record voice via sox, transcribe with whisper-cpp, normalize with
4
- // an OpenAI-compatible LLM, append to the TUI prompt (v1) or submit
5
- // immediately via session.prompt (v2 submit-only mode, see README
6
- // "OpenCode v1 and v2").
4
+ // an OpenAI-compatible LLM, and insert the text at the cursor of the
5
+ // focused input (composer, dialog search, or plugin field) on both
6
+ // hosts. Explicit submits only: /stt-submit, sttAutoSubmit, conversation.
7
7
  //
8
8
  // TTS: Auto-speak assistant responses (or read on demand) via Piper,
9
9
  // with LLM normalization for natural speech.
@@ -46,7 +46,8 @@ import { adaptV2Context } from "./lib/v2-compat.js";
46
46
  // Dual-compat export: v1's loader reads `tui`, v2's reads `setup`
47
47
  // (`Plugin.define` is an identity wrapper, so a plain object carries both).
48
48
  // Each host ignores the other's key. The core (`bootVoice`) is shared; v2
49
- // runs it over the v2-compat adapted api in submit-only mode.
49
+ // runs it over the v2-compat adapted api, which forwards the live renderer
50
+ // so dictation lands in whatever input currently has focus.
50
51
  export default {
51
52
  id: "opencode-voice",
52
53
  tui: async (api, options) => {
@@ -55,16 +56,16 @@ export default {
55
56
  setup: async (ctx) => {
56
57
  const adapted = adaptV2Context(ctx);
57
58
  await bootVoice(adapted.api, ctx.options, {
58
- submitOnly: true,
59
+ v2: true,
59
60
  onReady: async ({ logger }) => {
60
- // First-run submit-only notice (required UX disclosure): v2 has no
61
- // append-to-composer API, so every STT turn submits immediately.
61
+ // First-run v2 notice (required UX disclosure): text is typed into
62
+ // the focused field and is never sent automatically.
62
63
  const notice =
63
- "Voice on OpenCode v2: transcripts submit immediately as-is " +
64
- "(submit-only mode — append-for-edit is unavailable on v2). " +
65
- "For live preview, set sttMode to streaming.";
64
+ "Voice on OpenCode v2: transcripts are inserted at the cursor of the focused text field " +
65
+ "(prompt, search, or dialog input) and are not sent automatically. " +
66
+ "Set sttAutoSubmit to submit after recording.";
66
67
  adapted.api.ui.toast({ message: notice, variant: "warning", duration: 8000 });
67
- logger?.log?.("plugin", `v2 submit-only mode: ${notice}`, "warn");
68
+ logger?.log?.("plugin", `v2 focused-input mode: ${notice}`, "warn");
68
69
  // v2 has no client.app.log, so the logger above is a no-op: compat
69
70
  // warnings must surface as a toast or they are lost entirely.
70
71
  if (adapted.warnMessages.length > 0) {
package/lib/boot.js CHANGED
@@ -4,9 +4,9 @@
4
4
  // an `api` object (raw v1 `TuiPluginApi` vs the v2-compat adapted `Context`).
5
5
  // Everything below runs identically on both, so there is exactly one core to
6
6
  // maintain. The v1 `tui` path calls `bootVoice(api, options)` with no extras
7
- // and behaves exactly as before; the v2 `setup` path passes
8
- // `{ submitOnly: true }` for the first-run notice. No STT/TTS/routing/
9
- // keybind behavior branches live here.
7
+ // and behaves exactly as before; the v2 `setup` path passes `{ v2: true }`
8
+ // for the first-run notice. No STT/TTS/routing/keybind behavior branches
9
+ // live here.
10
10
 
11
11
  import fs from "node:fs";
12
12
  import os from "node:os";
@@ -98,7 +98,7 @@ export async function bootVoice(api, options, extras = {}) {
98
98
  ...cancel.commands,
99
99
  ]);
100
100
 
101
- if (extras?.submitOnly && typeof extras?.onReady === "function") {
101
+ if (extras?.v2 && typeof extras?.onReady === "function") {
102
102
  await extras.onReady({ logger, stt, tts, conversation, liveNotes, voiceModel, cancel });
103
103
  }
104
104
 
@@ -373,7 +373,7 @@ export function registerConversation(api, opts, logger, { stt, tts, isLiveNotesA
373
373
  // v2 prompt dispatch may emit the entire reply before its promise returns.
374
374
  // Home creation navigates before prompting, so resolve the ID at event time.
375
375
  let earlyWait = null;
376
- if (api.client?.tui?.submitOnly) {
376
+ if (api.client?.tui?.v2) {
377
377
  phase = "waiting";
378
378
  startReplyStream(currentSessionID, turnStart);
379
379
  earlyWait = waitForReply(currentSessionID);
package/lib/llm-client.js CHANGED
@@ -102,7 +102,9 @@ export function resolveProviderConfig(provider, model) {
102
102
  const exported = envVars.find((name) => name && process.env[name]);
103
103
  return {
104
104
  endpoint,
105
- model: model.id,
105
+ // v2 keeps the upstream provider's model id in `modelID` (`id` is the
106
+ // selection id and can differ for configured aliases), so prefer it.
107
+ model: model.modelID ?? model.id,
106
108
  apiKeyEnv: exported || envVars[0] || null,
107
109
  };
108
110
  }
package/lib/stt.js CHANGED
@@ -1376,9 +1376,11 @@ async function appendTranscription(client, renderer, text, submit, isCurrent) {
1376
1376
  return true;
1377
1377
  }
1378
1378
 
1379
- // Only the submit-only adapter consumes this fence; v1 request bodies
1380
- // remain unchanged. Re-check after an asynchronous home session create.
1381
- const fence = client.tui.submitOnly ? { isCurrent } : {};
1379
+ // Only the v2 adapter consumes this fence; v1 request bodies remain
1380
+ // unchanged. Re-check after an asynchronous home session create. The
1381
+ // `submit` flag lets the adapter distinguish an explicit submit from a
1382
+ // plain append with no editable field focused (append must not send).
1383
+ const fence = client.tui?.v2 ? { isCurrent, submit } : {};
1382
1384
  let appendResult = await client.tui.appendPrompt({ body: { text }, ...fence });
1383
1385
 
1384
1386
  if (appendResult?.error?.data?.message === "Expected object, got undefined") {
@@ -1391,7 +1393,9 @@ async function appendTranscription(client, renderer, text, submit, isCurrent) {
1391
1393
  );
1392
1394
  }
1393
1395
 
1394
- if (submit) {
1396
+ if (appendResult?.retained) return appendResult;
1397
+
1398
+ if (submit && !appendResult?.submitted) {
1395
1399
  await client.tui.submitPrompt();
1396
1400
  }
1397
1401
  return appendResult;
@@ -1515,15 +1519,15 @@ async function appendTurnText(client, renderer, toast, text, logger) {
1515
1519
  );
1516
1520
  clearProcessingToast();
1517
1521
  logger?.log("STT", `Turn appended chars=${text.length}`, "debug");
1518
- // Submit-only hosts (OpenCode v2: no append-to-composer API) submit via
1519
- // the client shim, which marks the result — so the toast must not claim
1520
- // the text was "added to prompt" for editing there. v1 results carry no
1522
+ // The v2 adapter retains the transcript when no editable field has
1523
+ // focus — it never sends an unrequested prompt. v1 results carry no
1521
1524
  // marker and keep the original toast.
1525
+ const retained = appendResult?.retained === true;
1522
1526
  toast(
1523
- appendResult?.submitted === true
1524
- ? "Transcription submitted as-is (submit-only mode)"
1527
+ retained
1528
+ ? "No text field focused - transcription kept (not sent)"
1525
1529
  : "Transcription added to prompt",
1526
- "success",
1530
+ retained ? "warning" : "success",
1527
1531
  );
1528
1532
  return { text };
1529
1533
  } catch (err) {
@@ -2136,7 +2140,12 @@ export function registerSTT(api, kv, complete, prompts, opts, logger, deps = {})
2136
2140
  cancel: () => cancelRecording(logger),
2137
2141
  discard: () => discardTurn(),
2138
2142
  transcribeTurn: () => transcribeTurn(kv, complete, client, api, toast, systemPrompt, logger),
2139
- submitTurnText: (text) => submitTurnText(client, renderer, toast, text, logger),
2143
+ // Conversation turns are hands-free submissions, not dictation: on v2
2144
+ // they must not press Enter in whatever field happens to have focus (a
2145
+ // permission dialog could own it), so they always go through the
2146
+ // session.prompt submit path. v1 keeps its focused-insert + submit.
2147
+ submitTurnText: (text) =>
2148
+ submitTurnText(client, client.tui?.v2 ? null : renderer, toast, text, logger),
2140
2149
  startStreaming: startStreamingDictation,
2141
2150
  finalizeStreaming: finalizeStreamingDictation,
2142
2151
  cancelStreaming: cancelStreamingDictation,
package/lib/v2-compat.js CHANGED
@@ -1,6 +1,6 @@
1
1
  import { createV2Preview } from "./v2-preview.js";
2
2
 
3
- // OpenCode v2 compatibility shim (submit-only mode).
3
+ // OpenCode v2 compatibility shim.
4
4
  //
5
5
  // WHY this file exists: v2's TUI plugin `Context` renamed or replaced every
6
6
  // v1 `TuiPluginApi` surface this plugin uses (see the drift table in the
@@ -10,12 +10,15 @@ import { createV2Preview } from "./v2-preview.js";
10
10
  // presence, never version strings), and every surface with no v2 path
11
11
  // degrades loudly (warn log + honest toast), never silently.
12
12
  //
13
- // Non-negotiable product decision (user-accepted): v2 2.0.26 has no
14
- // plugin-reachable append-to-composer API (`tui.prompt.append` exists only
15
- // as a server→TUI event type, not a client method). So on v2, STT
16
- // transcribes and SUBMITS immediately via `session.prompt`. The adapted api
17
- // passes `renderer: null` so `appendTranscription` skips the focused-input
18
- // insert path and lands on the client shim below, which submits.
13
+ // Composer editing on v2: there is no plugin-reachable append-to-composer
14
+ // call (`tui.prompt.append` exists only as a server→TUI event type, not a
15
+ // client method), but `Context.renderer` IS the live `CliRenderer`. So the
16
+ // adapted api forwards it, and the shared `insertIntoFocusedInput` helper
17
+ // types at the cursor of whatever editable field currently has focus — the
18
+ // session composer, a dialog search box, or another plugin's input. Only an
19
+ // explicit submit (`/stt-submit`, `sttAutoSubmit`, conversation) presses
20
+ // Enter. With no editable field focused, text is retained in a recovery
21
+ // dialog instead of being sent as a prompt.
19
22
 
20
23
  // Event renames with no same-named v2 equivalent. Values are the v2 event
21
24
  // type to subscribe to instead (both verified present in the v2
@@ -301,6 +304,20 @@ export function adaptV2Context(ctx) {
301
304
  return task;
302
305
  }
303
306
 
307
+ // Reached only when no focused renderable accepts text. Never redirect the
308
+ // transcript into a chat prompt the user did not ask for: keep it visible
309
+ // in a recovery dialog and let the caller report `retained`.
310
+ function retainTranscript(text) {
311
+ const captured = String(text || "");
312
+ try {
313
+ void ctx.ui.dialog.alert({
314
+ title: "Voice transcript — no text field focused",
315
+ message: captured,
316
+ });
317
+ } catch {}
318
+ return { data: null, retained: true };
319
+ }
320
+
304
321
  // `Object.create` (not spread) so client methods keep their `this`.
305
322
  // v2 `session.message` is a `{get}` namespace, not the v1 callable, so
306
323
  // the v1 call shape `client.session.message({sessionID, messageID})` is
@@ -323,14 +340,21 @@ export function adaptV2Context(ctx) {
323
340
  app: { value: { log: async () => {} } },
324
341
  tui: {
325
342
  value: {
326
- submitOnly: true,
327
- // Submit-only: append-for-edit is unimplementable on v2, so append
328
- // becomes submit. Accepts both v1 call shapes ({body:{text}} and
329
- // {text}) used by stt.js. The `submitted` marker lets the caller's
330
- // toast report a submit instead of an edit-step append.
343
+ // Marker for v1-shaped callers: this host has no composer-append API
344
+ // and dispatches prompts through `session.prompt` below (an explicit
345
+ // submit may resolve after the reply stream has started).
346
+ v2: true,
347
+ // Called only after `insertIntoFocusedInput` found no editable field.
348
+ // Accepts both v1 call shapes ({body:{text}} and {text}) used by
349
+ // stt.js. Explicit submits still go out via `session.prompt`; plain
350
+ // appends are retained visibly instead of being sent.
331
351
  appendPrompt: async (input) => {
332
- await submitTextAsPrompt(input?.body?.text ?? input?.text ?? "", input?.isCurrent);
333
- return { data: null, submitted: true };
352
+ const text = input?.body?.text ?? input?.text ?? "";
353
+ if (input?.submit === true) {
354
+ await submitTextAsPrompt(text, input?.isCurrent);
355
+ return { data: null, submitted: true };
356
+ }
357
+ return retainTranscript(text);
334
358
  },
335
359
  submitPrompt: async () => {},
336
360
  },
@@ -341,11 +365,11 @@ export function adaptV2Context(ctx) {
341
365
 
342
366
  const api = {
343
367
  client,
344
- // Submit-only enforcement: with no renderer, `appendTranscription`
345
- // skips `insertIntoFocusedInput` and lands on the client shim above,
346
- // which submits via `session.prompt`. Streaming display is separate:
368
+ // The live CliRenderer: `currentFocusedRenderable` lets the shared
369
+ // `insertIntoFocusedInput` helper type into the focused composer, dialog
370
+ // search box, or plugin input. Streaming display is separate:
347
371
  // voicePreview uses additive slots without composer editing.
348
- renderer: null,
372
+ renderer: ctx.renderer ?? null,
349
373
  voicePreview: preview,
350
374
  ui: { toast, dialog, DialogSelect },
351
375
  event: {
@@ -527,13 +551,44 @@ export function adaptV2Context(ctx) {
527
551
  },
528
552
  };
529
553
 
530
- // v1 reads `api.state.provider` as an array (`resolveVoiceProviderModel`
531
- // does `api?.state?.provider || []` then `.find`); v2 lists per location.
532
- // A getter keeps the array shape without touching callers.
554
+ // v1 reads `api.state.provider` as an array of `{models, env, options}`
555
+ // (`voice-model.js` / `llm-client.js`). v2 splits those across collections:
556
+ // providers carry NO models (they live in `location.model`), auth env names
557
+ // live on the provider's integration method `{type:"env", names}`, and the
558
+ // endpoint is `settings.baseURL` (`options` in v1). Recombining them here is
559
+ // what keeps /voice-model from seeing every provider as model-less.
533
560
  Object.defineProperty(api.state, "provider", {
534
561
  get() {
535
562
  try {
536
- return ctx.data.location.provider.list() || [];
563
+ const providers = ctx.data.location.provider.list() || [];
564
+ const models = ctx.data.location.model?.list?.() || [];
565
+ const integrations = ctx.data.location.integration?.list?.() || [];
566
+ const envByIntegration = new Map();
567
+ for (const integration of integrations) {
568
+ const names = (integration?.methods || [])
569
+ .filter((method) => method?.type === "env")
570
+ .flatMap((method) => method?.names || []);
571
+ if (names.length > 0) envByIntegration.set(integration.id, names);
572
+ }
573
+ const modelsByProvider = new Map();
574
+ for (const model of models) {
575
+ const providerID = model?.providerID;
576
+ if (!providerID) continue;
577
+ const list = modelsByProvider.get(providerID);
578
+ if (list) list.push(model);
579
+ else modelsByProvider.set(providerID, [model]);
580
+ }
581
+ return providers.map((provider) => {
582
+ const own = modelsByProvider.get(provider.id) || [];
583
+ return {
584
+ ...provider,
585
+ models: Object.fromEntries(own.map((model) => [model.id, model])),
586
+ env:
587
+ envByIntegration.get(provider.integrationID ?? provider.id) ??
588
+ (Array.isArray(provider.env) ? provider.env : []),
589
+ options: provider.settings || {},
590
+ };
591
+ });
537
592
  } catch {
538
593
  return [];
539
594
  }
@@ -554,7 +609,6 @@ export function adaptV2Context(ctx) {
554
609
  return {
555
610
  api,
556
611
  cleanup,
557
- submitOnly: true,
558
612
  warnMessages,
559
613
  onWarn: (sink) => void (warnSink = sink),
560
614
  };
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@bojackduy/opencode-voice",
3
- "version": "0.11.1",
3
+ "version": "0.11.2",
4
4
  "description": "Speech-to-text and text-to-speech for OpenCode. Record voice prompts with whisper-cpp, hear responses via Piper TTS, with LLM normalization through any OpenAI-compatible endpoint.",
5
5
  "keywords": [
6
6
  "opencode",