@framers/agentos-ext-voice-synthesis 2.0.0 → 2.0.1

package/README.md

@@ -1,6 +1,9 @@
 # Voice Synthesis Extension for AgentOS
 
-Text-to-speech synthesis using ElevenLabs with 8 predefined voices.
+Voice input/output tools for AgentOS:
+
+- `text_to_speech` via OpenAI, ElevenLabs, or a local Ollama-compatible runtime
+- `speech_to_text` via OpenAI Whisper, Deepgram, or Whisper-local/OpenAI-compatible local STT runtimes
 
 ## Installation
 
@@ -10,18 +13,73 @@ npm install @framers/agentos-ext-voice-synthesis
 
 ## Configuration
 
-Set `ELEVENLABS_API_KEY` environment variable or pass via options.
+Set one or more of the following environment variables, or pass them via
+extension options:
+
+- `OPENAI_API_KEY`
+- `ELEVENLABS_API_KEY`
+- `DEEPGRAM_API_KEY`
+- `OLLAMA_BASE_URL` for a local OpenAI-compatible TTS runtime
+- `WHISPER_LOCAL_BASE_URL` for a local OpenAI-compatible STT runtime
+- `TTS_PROVIDER` to prefer `openai`, `elevenlabs`, `ollama`, or `auto`
+- `STT_PROVIDER` to prefer `openai`, `deepgram`, `whisper-local`, or `auto`
+- `OPENAI_BASE_URL` for OpenAI-compatible speech endpoints
 
 ## Tool: text_to_speech
 
 **Input:**
-- `text` (string, required) — Text to convert (max 5000 chars)
-- `voice` (string, default: rachel) — Voice: rachel, domi, bella, antoni, josh, arnold, adam, sam
-- `model` (string, default: eleven_monolingual_v1) — ElevenLabs model
-- `stability` (number, 0-1, default: 0.5) — Voice stability
-- `similarity_boost` (number, 0-1, default: 0.75) — Voice similarity
+- `text` (string, required) — Text to convert, max 5000 chars
+- `provider` (string, optional) — `openai`, `elevenlabs`, `ollama`, or `auto`
+- `voice` (string, optional) — OpenAI: `alloy`, `echo`, `fable`, `onyx`, `nova`, `shimmer`; ElevenLabs: `rachel`, `domi`, `bella`, `antoni`, `josh`, `arnold`, `adam`, `sam`
+- `model` (string, optional) — OpenAI: `tts-1`, `tts-1-hd`; ElevenLabs: `eleven_monolingual_v1`, `eleven_multilingual_v2`
+- `speed` (number, OpenAI only) — 0.25 to 4.0
+- `stability` (number, ElevenLabs only) — 0 to 1
+- `similarity_boost` (number, ElevenLabs only) — 0 to 1
+- `format` (string) — `mp3`, `opus`, `aac`, `flac`, `wav`
+
+**Output:** Base64-encoded audio, provider metadata, and a duration estimate.
+
+## Provider selection
+
+The tool prefers providers in this order when `provider` is omitted:
+
+1. OpenAI, if `OPENAI_API_KEY` is set
+2. ElevenLabs, if `ELEVENLABS_API_KEY` is set
+3. Ollama-compatible local runtime as a best-effort fallback
+
+Ollama support is experimental and assumes an OpenAI-compatible TTS endpoint at
+`/v1/audio/speech`.
+
+## Tool: speech_to_text
+
+**Providers:**
+- `openai` — hosted Whisper via `OPENAI_API_KEY`
+- `deepgram` — hosted Deepgram STT via `DEEPGRAM_API_KEY`
+- `whisper-local` — local OpenAI-compatible transcription endpoint via `WHISPER_LOCAL_BASE_URL`
+- `auto` — prefers OpenAI, then Deepgram, then explicitly configured local STT
+
+**Input:**
+- `audioBase64` (string) — Base64 audio payload, optionally as a `data:` URL
+- `audioUrl` (string) — Fetchable audio URL
+- `provider` (string, optional) — `auto`, `openai`, `deepgram`, or `whisper-local`
+- `mimeType` (string, optional) — For example `audio/wav`
+- `fileName` (string, optional) — File name sent to the provider
+- `format` (string, optional) — Audio format hint such as `wav`, `mp3`, `m4a`, `webm`
+- `language` (string, optional) — ISO language hint
+- `prompt` (string, optional) — Context prompt to bias transcription
+- `model` (string, optional) — Provider model override such as `whisper-1`, `nova-2`, or `base`
+- `temperature` (number, optional) — Whisper temperature override
+- `responseFormat` (string, optional) — `json`, `text`, `srt`, `verbose_json`, or `vtt`
+- `diarize` (boolean, optional) — Enable speaker diarization where supported
+- `utterances` (boolean, optional) — Request utterance segmentation where supported
+- `smartFormat` (boolean, optional) — Enable provider-side formatting where supported
+- `detectLanguage` (boolean, optional) — Enable provider-side language detection where supported
+
+**Output:** Transcribed text, provider/model metadata, language, optional confidence, duration, and optional segments.
 
-**Output:** Base64-encoded MP3 audio with duration estimate.
+`whisper-local` targets OpenAI-compatible local transcription servers. That
+keeps the tool contract stable even when you swap between hosted Whisper,
+Deepgram, and local runtimes behind the same API shape.
 
 ## License
 

package/dist/index.d.ts

@@ -1,32 +1,45 @@
 /**
- * Voice Synthesis Extension Pack — multi-provider TTS for agents.
+ * Voice Synthesis Extension Pack — TTS + STT voice tools for agents.
  *
- * Supports: OpenAI TTS, ElevenLabs, Ollama (local).
+ * Supports: OpenAI TTS/STT, ElevenLabs TTS, Ollama (local TTS).
  * Auto-detects available provider from API keys.
  */
 import { TextToSpeechTool, type TTSProvider } from './tools/textToSpeech.js';
+import { SpeechToTextTool, type STTProvider } from './tools/speechToText.js';
 export interface VoiceSynthesisExtensionOptions {
     elevenLabsApiKey?: string;
     openaiApiKey?: string;
     openaiBaseUrl?: string;
+    deepgramApiKey?: string;
+    deepgramBaseUrl?: string;
     ollamaBaseUrl?: string;
+    whisperLocalBaseUrl?: string;
     defaultProvider?: TTSProvider;
+    defaultSttProvider?: STTProvider;
     priority?: number;
 }
 export declare function createExtensionPack(context: any): {
     name: string;
     version: string;
-    descriptors: {
+    descriptors: ({
         id: string;
         kind: "tool";
         priority: number;
         payload: TextToSpeechTool;
         requiredSecrets: never[];
-    }[];
+    } | {
+        id: string;
+        kind: "tool";
+        priority: number;
+        payload: SpeechToTextTool;
+        requiredSecrets: never[];
+    })[];
     onActivate: () => Promise<void>;
     onDeactivate: () => Promise<void>;
 };
 export { TextToSpeechTool };
+export { SpeechToTextTool };
 export type { TTSInput, TTSOutput, TTSConfig, TTSProvider } from './tools/textToSpeech.js';
+export type { STTInput, STTOutput, STTConfig, STTProvider } from './tools/speechToText.js';
 export default createExtensionPack;
 //# sourceMappingURL=index.d.ts.map

package/dist/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,EAAE,gBAAgB,EAAkB,KAAK,WAAW,EAAE,MAAM,yBAAyB,CAAC;AAE7F,MAAM,WAAW,8BAA8B;IAC7C,gBAAgB,CAAC,EAAE,MAAM,CAAC;IAC1B,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,eAAe,CAAC,EAAE,WAAW,CAAC;IAC9B,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,wBAAgB,mBAAmB,CAAC,OAAO,EAAE,GAAG;;;;;;;;;;;;EAsC/C;AAED,OAAO,EAAE,gBAAgB,EAAE,CAAC;AAC5B,YAAY,EAAE,QAAQ,EAAE,SAAS,EAAE,SAAS,EAAE,WAAW,EAAE,MAAM,yBAAyB,CAAC;AAC3F,eAAe,mBAAmB,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,EAAE,gBAAgB,EAAkB,KAAK,WAAW,EAAE,MAAM,yBAAyB,CAAC;AAC7F,OAAO,EAAE,gBAAgB,EAAkB,KAAK,WAAW,EAAE,MAAM,yBAAyB,CAAC;AAE7F,MAAM,WAAW,8BAA8B;IAC7C,gBAAgB,CAAC,EAAE,MAAM,CAAC;IAC1B,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,mBAAmB,CAAC,EAAE,MAAM,CAAC;IAC7B,eAAe,CAAC,EAAE,WAAW,CAAC;IAC9B,kBAAkB,CAAC,EAAE,WAAW,CAAC;IACjC,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,wBAAgB,mBAAmB,CAAC,OAAO,EAAE,GAAG;;;;;;;;;;;;;;;;;;EA8D/C;AAED,OAAO,EAAE,gBAAgB,EAAE,CAAC;AAC5B,OAAO,EAAE,gBAAgB,EAAE,CAAC;AAC5B,YAAY,EAAE,QAAQ,EAAE,SAAS,EAAE,SAAS,EAAE,WAAW,EAAE,MAAM,yBAAyB,CAAC;AAC3F,YAAY,EAAE,QAAQ,EAAE,SAAS,EAAE,SAAS,EAAE,WAAW,EAAE,MAAM,yBAAyB,CAAC;AAC3F,eAAe,mBAAmB,CAAC"}

package/dist/index.js

@@ -1,10 +1,11 @@
 /**
- * Voice Synthesis Extension Pack — multi-provider TTS for agents.
+ * Voice Synthesis Extension Pack — TTS + STT voice tools for agents.
  *
- * Supports: OpenAI TTS, ElevenLabs, Ollama (local).
+ * Supports: OpenAI TTS/STT, ElevenLabs TTS, Ollama (local TTS).
  * Auto-detects available provider from API keys.
  */
 import { TextToSpeechTool } from './tools/textToSpeech.js';
+import { SpeechToTextTool } from './tools/speechToText.js';
 export function createExtensionPack(context) {
     const options = (context.options || {});
     const config = {
@@ -14,28 +15,54 @@ export function createExtensionPack(context) {
         ollamaBaseUrl: options.ollamaBaseUrl || process.env.OLLAMA_BASE_URL,
         defaultProvider: options.defaultProvider || process.env.TTS_PROVIDER || 'auto',
     };
-    const tool = new TextToSpeechTool(config);
-    // Determine which providers are available for the activation message
+    const sttConfig = {
+        openaiApiKey: options.openaiApiKey || context.getSecret?.('openai.apiKey') || process.env.OPENAI_API_KEY,
+        openaiBaseUrl: options.openaiBaseUrl || process.env.OPENAI_BASE_URL,
+        deepgramApiKey: options.deepgramApiKey || context.getSecret?.('deepgram.apiKey') || process.env.DEEPGRAM_API_KEY,
+        deepgramBaseUrl: options.deepgramBaseUrl || process.env.DEEPGRAM_BASE_URL,
+        whisperLocalBaseUrl: options.whisperLocalBaseUrl || process.env.WHISPER_LOCAL_BASE_URL,
+        defaultProvider: options.defaultSttProvider || process.env.STT_PROVIDER || 'auto',
+    };
+    const ttsTool = new TextToSpeechTool(config);
+    const sttTool = new SpeechToTextTool(sttConfig);
     const providers = [];
     if (config.openaiApiKey)
         providers.push('OpenAI');
     if (config.elevenLabsApiKey)
         providers.push('ElevenLabs');
     providers.push('Ollama (local fallback)');
+    const sttProviders = [];
+    if (sttConfig.openaiApiKey)
+        sttProviders.push('OpenAI Whisper');
+    if (sttConfig.deepgramApiKey)
+        sttProviders.push('Deepgram');
+    if (sttConfig.whisperLocalBaseUrl)
+        sttProviders.push('Whisper-local');
+    const features = [
+        `TTS: ${providers.join(', ')}`,
+        `STT: ${sttProviders.length > 0 ? sttProviders.join(', ') : 'not configured'}`,
+    ];
     return {
         name: '@framers/agentos-ext-voice-synthesis',
         version: '2.0.0',
         descriptors: [
             {
-                id: tool.name,
+                id: ttsTool.name,
+                kind: 'tool',
+                priority: options.priority || 50,
+                payload: ttsTool,
+                requiredSecrets: [],
+            },
+            {
+                id: sttTool.name,
                 kind: 'tool',
                 priority: options.priority || 50,
-                payload: tool,
+                payload: sttTool,
                 requiredSecrets: [],
             },
         ],
         onActivate: async () => {
-            context.logger?.info?.(`Voice Synthesis activated — providers: ${providers.join(', ')}`);
+            context.logger?.info?.(`Voice Synthesis activated — ${features.join(' | ')}`);
         },
         onDeactivate: async () => {
             context.logger?.info?.('Voice Synthesis deactivated');
@@ -43,5 +70,6 @@ export function createExtensionPack(context) {
     };
 }
 export { TextToSpeechTool };
+export { SpeechToTextTool };
 export default createExtensionPack;
 //# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -1 +1 @@
-{"version":3,"file":"index.js","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,EAAE,gBAAgB,EAAoC,MAAM,yBAAyB,CAAC;AAW7F,MAAM,UAAU,mBAAmB,CAAC,OAAY;IAC9C,MAAM,OAAO,GAAG,CAAC,OAAO,CAAC,OAAO,IAAI,EAAE,CAAmC,CAAC;IAE1E,MAAM,MAAM,GAAc;QACxB,YAAY,EAAE,OAAO,CAAC,YAAY,IAAI,OAAO,CAAC,SAAS,EAAE,CAAC,eAAe,CAAC,IAAI,OAAO,CAAC,GAAG,CAAC,cAAc;QACxG,aAAa,EAAE,OAAO,CAAC,aAAa,IAAI,OAAO,CAAC,GAAG,CAAC,eAAe;QACnE,gBAAgB,EAAE,OAAO,CAAC,gBAAgB,IAAI,OAAO,CAAC,SAAS,EAAE,CAAC,mBAAmB,CAAC,IAAI,OAAO,CAAC,GAAG,CAAC,kBAAkB;QACxH,aAAa,EAAE,OAAO,CAAC,aAAa,IAAI,OAAO,CAAC,GAAG,CAAC,eAAe;QACnE,eAAe,EAAE,OAAO,CAAC,eAAe,IAAK,OAAO,CAAC,GAAG,CAAC,YAA4B,IAAI,MAAM;KAChG,CAAC;IAEF,MAAM,IAAI,GAAG,IAAI,gBAAgB,CAAC,MAAM,CAAC,CAAC;IAE1C,qEAAqE;IACrE,MAAM,SAAS,GAAa,EAAE,CAAC;IAC/B,IAAI,MAAM,CAAC,YAAY;QAAE,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IAClD,IAAI,MAAM,CAAC,gBAAgB;QAAE,SAAS,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC;IAC1D,SAAS,CAAC,IAAI,CAAC,yBAAyB,CAAC,CAAC;IAE1C,OAAO;QACL,IAAI,EAAE,sCAAsC;QAC5C,OAAO,EAAE,OAAO;QAChB,WAAW,EAAE;YACX;gBACE,EAAE,EAAE,IAAI,CAAC,IAAI;gBACb,IAAI,EAAE,MAAe;gBACrB,QAAQ,EAAE,OAAO,CAAC,QAAQ,IAAI,EAAE;gBAChC,OAAO,EAAE,IAAI;gBACb,eAAe,EAAE,EAAE;aACpB;SACF;QACD,UAAU,EAAE,KAAK,IAAI,EAAE;YACrB,OAAO,CAAC,MAAM,EAAE,IAAI,EAAE,CAAC,0CAA0C,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QAC3F,CAAC;QACD,YAAY,EAAE,KAAK,IAAI,EAAE;YACvB,OAAO,CAAC,MAAM,EAAE,IAAI,EAAE,CAAC,6BAA6B,CAAC,CAAC;QACxD,CAAC;KACF,CAAC;AACJ,CAAC;AAED,OAAO,EAAE,gBAAgB,EAAE,CAAC;AAE5B,eAAe,mBAAmB,CAAC"}
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,EAAE,gBAAgB,EAAoC,MAAM,yBAAyB,CAAC;AAC7F,OAAO,EAAE,gBAAgB,EAAoC,MAAM,yBAAyB,CAAC;AAe7F,MAAM,UAAU,mBAAmB,CAAC,OAAY;IAC9C,MAAM,OAAO,GAAG,CAAC,OAAO,CAAC,OAAO,IAAI,EAAE,CAAmC,CAAC;IAE1E,MAAM,MAAM,GAAc;QACxB,YAAY,EAAE,OAAO,CAAC,YAAY,IAAI,OAAO,CAAC,SAAS,EAAE,CAAC,eAAe,CAAC,IAAI,OAAO,CAAC,GAAG,CAAC,cAAc;QACxG,aAAa,EAAE,OAAO,CAAC,aAAa,IAAI,OAAO,CAAC,GAAG,CAAC,eAAe;QACnE,gBAAgB,EAAE,OAAO,CAAC,gBAAgB,IAAI,OAAO,CAAC,SAAS,EAAE,CAAC,mBAAmB,CAAC,IAAI,OAAO,CAAC,GAAG,CAAC,kBAAkB;QACxH,aAAa,EAAE,OAAO,CAAC,aAAa,IAAI,OAAO,CAAC,GAAG,CAAC,eAAe;QACnE,eAAe,EAAE,OAAO,CAAC,eAAe,IAAK,OAAO,CAAC,GAAG,CAAC,YAA4B,IAAI,MAAM;KAChG,CAAC;IACF,MAAM,SAAS,GAAc;QAC3B,YAAY,EAAE,OAAO,CAAC,YAAY,IAAI,OAAO,CAAC,SAAS,EAAE,CAAC,eAAe,CAAC,IAAI,OAAO,CAAC,GAAG,CAAC,cAAc;QACxG,aAAa,EAAE,OAAO,CAAC,aAAa,IAAI,OAAO,CAAC,GAAG,CAAC,eAAe;QACnE,cAAc,EAAE,OAAO,CAAC,cAAc,IAAI,OAAO,CAAC,SAAS,EAAE,CAAC,iBAAiB,CAAC,IAAI,OAAO,CAAC,GAAG,CAAC,gBAAgB;QAChH,eAAe,EAAE,OAAO,CAAC,eAAe,IAAI,OAAO,CAAC,GAAG,CAAC,iBAAiB;QACzE,mBAAmB,EAAE,OAAO,CAAC,mBAAmB,IAAI,OAAO,CAAC,GAAG,CAAC,sBAAsB;QACtF,eAAe,EAAE,OAAO,CAAC,kBAAkB,IAAK,OAAO,CAAC,GAAG,CAAC,YAA4B,IAAI,MAAM;KACnG,CAAC;IAEF,MAAM,OAAO,GAAG,IAAI,gBAAgB,CAAC,MAAM,CAAC,CAAC;IAC7C,MAAM,OAAO,GAAG,IAAI,gBAAgB,CAAC,SAAS,CAAC,CAAC;IAEhD,MAAM,SAAS,GAAa,EAAE,CAAC;IAC/B,IAAI,MAAM,CAAC,YAAY;QAAE,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IAClD,IAAI,MAAM,CAAC,gBAAgB;QAAE,SAAS,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC;IAC1D,SAAS,CAAC,IAAI,CAAC,yBAAyB,CAAC,CAAC;IAC1C,MAAM,YAAY,GAAa,EAAE,CAAC;IAClC,IAAI,SAAS,CAAC,YAAY;QAAE,YAAY,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAC;IAChE,IAAI,SAAS,CAAC,cAAc;QAAE,YAAY,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;IAC5D,IAAI,SAAS,CAAC,mBAAmB;QAAE,YAAY,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC;IAEtE,MAAM,QAAQ,GAAa;QACzB,QAAQ,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE;QAC9B,QAAQ,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,gBAAgB,EAAE;KAC/E,CAAC;IAEF,OAAO;QACL,IAAI,EAAE,sCAAsC;QAC5C,OAAO,EAAE,OAAO;QAChB,WAAW,EAAE;YACX;gBACE,EAAE,EAAE,OAAO,CAAC,IAAI;gBAChB,IAAI,EAAE,MAAe;gBACrB,QAAQ,EAAE,OAAO,CAAC,QAAQ,IAAI,EAAE;gBAChC,OAAO,EAAE,OAAO;gBAChB,eAAe,EAAE,EAAE;aACpB;YACD;gBACE,EAAE,EAAE,OAAO,CAAC,IAAI;gBAChB,IAAI,EAAE,MAAe;gBACrB,QAAQ,EAAE,OAAO,CAAC,QAAQ,IAAI,EAAE;gBAChC,OAAO,EAAE,OAAO;gBAChB,eAAe,EAAE,EAAE;aACpB;SACF;QACD,UAAU,EAAE,KAAK,IAAI,EAAE;YACrB,OAAO,CAAC,MAAM,EAAE,IAAI,EAAE,CAAC,+BAA+B,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;QAChF,CAAC;QACD,YAAY,EAAE,KAAK,IAAI,EAAE;YACvB,OAAO,CAAC,MAAM,EAAE,IAAI,EAAE,CAAC,6BAA6B,CAAC,CAAC;QACxD,CAAC;KACF,CAAC;AACJ,CAAC;AAED,OAAO,EAAE,gBAAgB,EAAE,CAAC;AAC5B,OAAO,EAAE,gBAAgB,EAAE,CAAC;AAG5B,eAAe,mBAAmB,CAAC"}

package/dist/tools/speechToText.d.ts

@@ -0,0 +1,77 @@
+/**
+ * Multi-provider STT Tool — speech-to-text transcription.
+ *
+ * Supports: OpenAI Whisper, Deepgram, and Whisper-local/OpenAI-compatible
+ * local runtimes behind one stable tool contract.
+ */
+import type { ITool, JSONSchemaObject, ToolExecutionContext, ToolExecutionResult } from '@framers/agentos';
+export type STTProvider = 'openai' | 'deepgram' | 'whisper-local' | 'auto';
+export type STTResponseFormat = 'json' | 'text' | 'srt' | 'verbose_json' | 'vtt';
+export interface STTInput {
+    audioBase64?: string;
+    audioUrl?: string;
+    mimeType?: string;
+    fileName?: string;
+    format?: string;
+    language?: string;
+    prompt?: string;
+    temperature?: number;
+    responseFormat?: STTResponseFormat;
+    provider?: STTProvider;
+    model?: string;
+    diarize?: boolean;
+    utterances?: boolean;
+    smartFormat?: boolean;
+    detectLanguage?: boolean;
+    providerOptions?: Record<string, unknown>;
+}
+export interface STTWord {
+    word: string;
+    start: number;
+    end: number;
+    confidence?: number;
+    speaker?: string | number;
+}
+export interface STTSegment {
+    text: string;
+    start: number;
+    end: number;
+    confidence?: number;
+    speaker?: string | number;
+    words?: STTWord[];
+}
+export interface STTOutput {
+    text: string;
+    provider: Exclude<STTProvider, 'auto'>;
+    model: string;
+    language?: string;
+    confidence?: number;
+    durationSeconds?: number;
+    segments?: STTSegment[];
+    providerResponse?: unknown;
+}
+export interface STTConfig {
+    openaiApiKey?: string;
+    openaiBaseUrl?: string;
+    deepgramApiKey?: string;
+    deepgramBaseUrl?: string;
+    whisperLocalBaseUrl?: string;
+    defaultProvider?: STTProvider;
+}
+export declare class SpeechToTextTool implements ITool<STTInput, STTOutput> {
+    readonly id = "stt-multi-provider-v1";
+    readonly name = "speech_to_text";
+    readonly displayName = "Speech to Text";
+    readonly description: string;
+    readonly category = "media";
+    readonly version = "2.0.0";
+    readonly hasSideEffects = false;
+    readonly requiredCapabilities: string[];
+    readonly inputSchema: JSONSchemaObject;
+    private readonly config;
+    constructor(config?: STTConfig);
+    private resolveProvider;
+    private prepareAudio;
+    execute(args: STTInput, _context: ToolExecutionContext): Promise<ToolExecutionResult<STTOutput>>;
+}
+//# sourceMappingURL=speechToText.d.ts.map

package/dist/tools/speechToText.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"speechToText.d.ts","sourceRoot":"","sources":["../../src/tools/speechToText.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,KAAK,EACV,KAAK,EACL,gBAAgB,EAChB,oBAAoB,EACpB,mBAAmB,EACpB,MAAM,kBAAkB,CAAC;AAE1B,MAAM,MAAM,WAAW,GAAG,QAAQ,GAAG,UAAU,GAAG,eAAe,GAAG,MAAM,CAAC;AAC3E,MAAM,MAAM,iBAAiB,GAAG,MAAM,GAAG,MAAM,GAAG,KAAK,GAAG,cAAc,GAAG,KAAK,CAAC;AAEjF,MAAM,WAAW,QAAQ;IACvB,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,cAAc,CAAC,EAAE,iBAAiB,CAAC;IACnC,QAAQ,CAAC,EAAE,WAAW,CAAC;IACvB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB,UAAU,CAAC,EAAE,OAAO,CAAC;IACrB,WAAW,CAAC,EAAE,OAAO,CAAC;IACtB,cAAc,CAAC,EAAE,OAAO,CAAC;IACzB,eAAe,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;CAC3C;AAED,MAAM,WAAW,OAAO;IACtB,IAAI,EAAE,MAAM,CAAC;IACb,KAAK,EAAE,MAAM,CAAC;IACd,GAAG,EAAE,MAAM,CAAC;IACZ,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,OAAO,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;CAC3B;AAED,MAAM,WAAW,UAAU;IACzB,IAAI,EAAE,MAAM,CAAC;IACb,KAAK,EAAE,MAAM,CAAC;IACd,GAAG,EAAE,MAAM,CAAC;IACZ,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,OAAO,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;IAC1B,KAAK,CAAC,EAAE,OAAO,EAAE,CAAC;CACnB;AAED,MAAM,WAAW,SAAS;IACxB,IAAI,EAAE,MAAM,CAAC;IACb,QAAQ,EAAE,OAAO,CAAC,WAAW,EAAE,MAAM,CAAC,CAAC;IACvC,KAAK,EAAE,MAAM,CAAC;IACd,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,QAAQ,CAAC,EAAE,UAAU,EAAE,CAAC;IACxB,gBAAgB,CAAC,EAAE,OAAO,CAAC;CAC5B;AAED,MAAM,WAAW,SAAS;IACxB,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,mBAAmB,CAAC,EAAE,MAAM,CAAC;IAC7B,eAAe,CAAC,EAAE,WAAW,CAAC;CAC/B;AA8YD,qBAAa,gBAAiB,YAAW,KAAK,CAAC,QAAQ,EAAE,SAAS,CAAC;IACjE,QAAQ,CAAC,EAAE,2BAA2B;IACtC,QAAQ,CAAC,IAAI,oBAAoB;IACjC,QAAQ,CAAC,WAAW,oBAAoB;IACxC,QAAQ,CAAC,WAAW,SAEsC;IAC1D,QAAQ,CAAC,QAAQ,WAAW;IAC5B,QAAQ,CAAC,OAAO,WAAW;IAC3B,QAAQ,CAAC,cAAc,SAAS;IAChC,QAAQ,CAAC,oBAAoB,WAAsB;IAEnD,QAAQ,CAAC,WAAW,EAAE,gBAAgB,CA0EpC;IAEF,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAoB;gBAE/B,MAAM,CAAC,EAAE,SAAS;IAgB9B,OAAO,CAAC,eAAe;YAcT,YAAY;IAgCpB,OAAO,CACX,IAAI,EAAE,QAAQ,EACd,QAAQ,EAAE,oBAAoB,GAC7B,OAAO,CAAC,mBAAmB,CAAC,SAAS,CAAC,CAAC;CAyB3C"}