@mastra/voice-inworld 0.0.0

package/README.md

@@ -0,0 +1,111 @@
+# @mastra/voice-inworld
+
+[Inworld AI](https://inworld.ai) voice provider for [Mastra](https://mastra.ai) — streaming TTS and batch STT.
+
+## Installation
+
+```bash
+npm install @mastra/voice-inworld @mastra/core
+```
+
+## Quick Start
+
+```typescript
+import { InworldVoice } from '@mastra/voice-inworld';
+
+const voice = new InworldVoice({
+  speaker: 'Dennis', // 22 built-in voices available
+});
+
+// Text-to-Speech (streaming)
+const audioStream = await voice.speak('Hello from Inworld!');
+
+// Speech-to-Text
+const transcript = await voice.listen(audioStream);
+
+// List available voices
+const voices = await voice.getSpeakers();
+```
+
+## Configuration
+
+```typescript
+const voice = new InworldVoice({
+  speechModel: {
+    name: 'inworld-tts-2', // default; also 'inworld-tts-1.5-max' or 'inworld-tts-1.5-mini'
+    apiKey: 'your-key', // or set INWORLD_API_KEY env var
+  },
+  listeningModel: {
+    name: 'groq/whisper-large-v3',
+  },
+  speaker: 'Dennis', // default voice
+  audioEncoding: 'MP3', // MP3, WAV, OGG_OPUS, LINEAR16, PCM, ALAW, MULAW, FLAC
+  sampleRateHertz: 48000, // 8000-48000
+  language: 'en-US', // BCP-47 language code for STT
+});
+```
+
+## Speak Options
+
+```typescript
+const stream = await voice.speak('Hello', {
+  speaker: 'Olivia', // override voice
+  audioEncoding: 'WAV', // override format
+  sampleRateHertz: 24000, // override sample rate
+  speakingRate: 1.2, // 0.5 - 1.5
+  temperature: 0.8, // (0, 2] — ignored on inworld-tts-2
+  deliveryMode: 'CREATIVE', // STABLE | BALANCED | CREATIVE — only honored on inworld-tts-2
+  language: 'fr-FR', // BCP-47 per-call override; auto-detected when omitted
+});
+```
+
+## Listen Options
+
+```typescript
+const text = await voice.listen(audioStream, {
+  audioEncoding: 'AUTO_DETECT', // or 'MP3', 'LINEAR16', etc.
+  sampleRateHertz: 16000,
+  language: 'en-US',
+});
+```
+
+## CompositeVoice
+
+Mix Inworld with other providers:
+
+```typescript
+import { CompositeVoice } from '@mastra/core/voice';
+import { InworldVoice } from '@mastra/voice-inworld';
+import { DeepgramVoice } from '@mastra/voice-deepgram';
+
+const voice = new CompositeVoice({
+  output: new InworldVoice({ speaker: 'Olivia' }), // Inworld for TTS
+  input: new DeepgramVoice(), // Deepgram for STT
+});
+```
+
+## Available Voices
+
+Alex, Ashley, Craig, Deborah, Dennis, Dominus, Edward, Elizabeth, Hades, Heitor, Julia, Maite, Mark, Olivia, Pixie, Priya, Ronald, Sarah, Shaun, Theodore, Timothy, Wendy.
+
+## TTS Models
+
+| Model                  | Quality | Latency       | Notes                                           |
+| ---------------------- | ------- | ------------- | ----------------------------------------------- |
+| `inworld-tts-2`        | Highest | ~200ms median | **Default.** Flagship; supports `deliveryMode`. |
+| `inworld-tts-1.5-max`  | High    | ~200ms median | Previous flagship. Supports `temperature`.      |
+| `inworld-tts-1.5-mini` | Good    | ~100ms median | Lower latency, reduced quality.                 |
+
+## STT Models
+
+| Model                   | Languages | Notes                      |
+| ----------------------- | --------- | -------------------------- |
+| `groq/whisper-large-v3` | 99+       | Best multilingual coverage |
+
+## Streaming
+
+The `speak()` method uses Inworld's streaming TTS endpoint (`/tts/v1/voice:stream`), returning audio chunks progressively as they are generated. This is ideal for agentic workflows where low time-to-first-audio matters.
+
+## Authentication
+
+Set your API key via the `INWORLD_API_KEY` environment variable or pass it in the config. Get your key from [platform.inworld.ai](https://platform.inworld.ai) → Settings → API Keys.

package/dist/index.cjs

@@ -0,0 +1,256 @@
+'use strict';
+
+var stream = require('stream');
+var voice = require('@mastra/core/voice');
+
+// src/index.ts
+var INWORLD_API_BASE = "https://api.inworld.ai";
+var InworldVoice = class extends voice.MastraVoice {
+  speechApiKey;
+  listeningApiKey;
+  audioEncoding;
+  sampleRateHertz;
+  language;
+  /**
+   * Creates an instance of the InworldVoice class.
+   *
+   * @param {Object} options - The options for the voice configuration.
+   * @param {InworldVoiceConfig} [options.speechModel] - TTS model config. Default: inworld-tts-2.
+   * @param {InworldListeningConfig} [options.listeningModel] - STT model config. Default: groq/whisper-large-v3.
+   * @param {string} [options.speaker] - Default voice ID. Default: 'Dennis'.
+   * @param {AudioEncoding} [options.audioEncoding] - TTS audio format. Default: 'MP3'.
+   * @param {number} [options.sampleRateHertz] - TTS sample rate. Default: 48000.
+   * @param {string} [options.language] - STT language (BCP-47). Default: 'en-US'.
+   *
+   * @throws {Error} If no API key is provided or found in INWORLD_API_KEY env var.
+   */
+  constructor({
+    speechModel,
+    listeningModel,
+    speaker,
+    audioEncoding,
+    sampleRateHertz,
+    language
+  } = {}) {
+    const envKey = process.env.INWORLD_API_KEY;
+    const speechApiKey = speechModel?.apiKey ?? listeningModel?.apiKey ?? envKey;
+    const listeningApiKey = listeningModel?.apiKey ?? speechModel?.apiKey ?? envKey;
+    if (!speechApiKey || !listeningApiKey) {
+      throw new Error(
+        "Inworld API key is required. Pass apiKey in speechModel/listeningModel config or set INWORLD_API_KEY env var."
+      );
+    }
+    super({
+      speechModel: {
+        name: speechModel?.name ?? "inworld-tts-2",
+        apiKey: speechApiKey
+      },
+      listeningModel: {
+        name: listeningModel?.name ?? "groq/whisper-large-v3",
+        apiKey: listeningApiKey
+      },
+      speaker: speaker ?? "Dennis"
+    });
+    this.speechApiKey = speechApiKey;
+    this.listeningApiKey = listeningApiKey;
+    this.audioEncoding = audioEncoding ?? "MP3";
+    this.sampleRateHertz = sampleRateHertz ?? 48e3;
+    this.language = language ?? "en-US";
+  }
+  /**
+   * Retrieves a list of available voices from the Inworld API.
+   *
+   * @returns {Promise<Array<{ voiceId: string; name: string; language: string; description: string; tags: string[]; source: string }>>}
+   */
+  async getSpeakers() {
+    const response = await fetch(`${INWORLD_API_BASE}/voices/v1/voices`, {
+      headers: { Authorization: `Basic ${this.speechApiKey}` }
+    });
+    if (!response.ok) {
+      const text = await response.text();
+      throw new Error(`Inworld list voices failed (${response.status}): ${text}`);
+    }
+    const data = await response.json();
+    return data.voices?.map((v) => ({
+      voiceId: v.voiceId,
+      name: v.displayName ?? v.voiceId,
+      language: v.langCode ?? "en",
+      description: v.description ?? "",
+      tags: v.tags ?? [],
+      source: v.source ?? "SYSTEM"
+    })) ?? [];
+  }
+  /**
+   * Checks if listening capabilities are enabled.
+   */
+  async getListener() {
+    return { enabled: !!this.listeningModel };
+  }
+  /**
+   * Converts text to speech using Inworld's streaming TTS endpoint.
+   *
+   * Returns a ReadableStream that emits audio chunks progressively as they
+   * arrive from the API — following the same streaming pattern used by the
+   * Deepgram and PlayAI Mastra voice providers.
+   *
+   * Uses the `/tts/v1/voice:stream` endpoint which returns newline-delimited
+   * JSON, each line containing a base64-encoded audio chunk.
+   *
+   * @param {string | NodeJS.ReadableStream} input - Text or stream to convert to speech.
+   * @param {InworldSpeakOptions} [options] - TTS options.
+   * @returns {Promise<NodeJS.ReadableStream>} Progressive audio stream.
+   */
+  async speak(input, options) {
+    const text = typeof input === "string" ? input : await this.streamToString(input);
+    if (text.trim().length === 0) {
+      throw new Error("Input text is empty");
+    }
+    const speaker = options?.speaker ?? this.speaker;
+    const body = {
+      text,
+      voiceId: speaker,
+      modelId: this.speechModel?.name ?? "inworld-tts-2",
+      audioConfig: {
+        audioEncoding: options?.audioEncoding ?? this.audioEncoding,
+        sampleRateHertz: options?.sampleRateHertz ?? this.sampleRateHertz,
+        ...options?.speakingRate !== void 0 && { speakingRate: options.speakingRate }
+      },
+      ...options?.temperature !== void 0 && { temperature: options.temperature },
+      ...options?.deliveryMode !== void 0 && { deliveryMode: options.deliveryMode },
+      ...options?.language !== void 0 && { language: options.language }
+    };
+    const response = await fetch(`${INWORLD_API_BASE}/tts/v1/voice:stream`, {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+        Authorization: `Basic ${this.speechApiKey}`
+      },
+      body: JSON.stringify(body)
+    });
+    if (!response.ok) {
+      const errorText = await response.text();
+      throw new Error(`Inworld TTS failed (${response.status}): ${errorText}`);
+    }
+    if (!response.body) {
+      throw new Error("Inworld TTS streaming response has no body");
+    }
+    const outputStream = new stream.PassThrough();
+    const reader = response.body.getReader();
+    const decoder = new TextDecoder();
+    let buffer = "";
+    const writeChunk = async (data) => {
+      if (outputStream.destroyed) return;
+      if (!outputStream.write(data)) {
+        await new Promise((resolve, reject) => {
+          outputStream.once("drain", resolve);
+          outputStream.once("close", resolve);
+          outputStream.once("error", reject);
+        });
+      }
+    };
+    outputStream.once("close", () => {
+      void reader.cancel().catch(() => {
+      });
+    });
+    void (async () => {
+      try {
+        while (true) {
+          if (outputStream.destroyed) {
+            void reader.cancel().catch(() => {
+            });
+            break;
+          }
+          const { done, value } = await reader.read();
+          if (done) {
+            const remaining = buffer.trim();
+            if (remaining) {
+              const chunk = JSON.parse(remaining);
+              const audioContent = chunk.result?.audioContent ?? chunk.audioContent;
+              if (audioContent) {
+                await writeChunk(Buffer.from(audioContent, "base64"));
+              }
+            }
+            outputStream.end();
+            break;
+          }
+          buffer += decoder.decode(value, { stream: true });
+          const lines = buffer.split("\n");
+          buffer = lines.pop() ?? "";
+          for (const line of lines) {
+            const trimmed = line.trim();
+            if (!trimmed) continue;
+            if (outputStream.destroyed) {
+              void reader.cancel().catch(() => {
+              });
+              return;
+            }
+            const chunk = JSON.parse(trimmed);
+            const audioContent = chunk.result?.audioContent ?? chunk.audioContent;
+            if (audioContent) {
+              await writeChunk(Buffer.from(audioContent, "base64"));
+            }
+          }
+        }
+      } catch (err) {
+        if (!outputStream.destroyed) {
+          outputStream.destroy(err instanceof Error ? err : new Error(String(err)));
+        }
+        void reader.cancel().catch(() => {
+        });
+      }
+    })().catch(() => {
+    });
+    return outputStream;
+  }
+  /**
+   * Converts audio to text using Inworld's batch STT endpoint.
+   *
+   * @param {NodeJS.ReadableStream} input - Audio stream to transcribe.
+   * @param {InworldListenOptions} [options] - STT options.
+   * @returns {Promise<string>} Transcribed text.
+   */
+  async listen(input, options) {
+    const chunks = [];
+    for await (const chunk of input) {
+      chunks.push(typeof chunk === "string" ? Buffer.from(chunk) : chunk);
+    }
+    const audioBase64 = Buffer.concat(chunks).toString("base64");
+    const body = {
+      transcribeConfig: {
+        modelId: this.listeningModel?.name ?? "groq/whisper-large-v3",
+        audioEncoding: options?.audioEncoding ?? "AUTO_DETECT",
+        language: options?.language ?? this.language,
+        sampleRateHertz: options?.sampleRateHertz ?? 16e3,
+        numberOfChannels: options?.numberOfChannels ?? 1
+      },
+      audioData: {
+        content: audioBase64
+      }
+    };
+    const response = await fetch(`${INWORLD_API_BASE}/stt/v1/transcribe`, {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+        Authorization: `Basic ${this.listeningApiKey}`
+      },
+      body: JSON.stringify(body)
+    });
+    if (!response.ok) {
+      const text = await response.text();
+      throw new Error(`Inworld STT failed (${response.status}): ${text}`);
+    }
+    const result = await response.json();
+    return result.transcription?.transcript ?? "";
+  }
+  async streamToString(stream) {
+    const chunks = [];
+    for await (const chunk of stream) {
+      chunks.push(typeof chunk === "string" ? Buffer.from(chunk) : chunk);
+    }
+    return Buffer.concat(chunks).toString("utf-8");
+  }
+};
+
+exports.InworldVoice = InworldVoice;
+//# sourceMappingURL=index.cjs.map
+//# sourceMappingURL=index.cjs.map

package/dist/index.cjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/index.ts"],"names":["MastraVoice","PassThrough"],"mappings":";;;;;;AAGA,IAAM,gBAAA,GAAmB,wBAAA;AAmDlB,IAAM,YAAA,GAAN,cAA2BA,iBAAA,CAAY;AAAA,EACpC,YAAA;AAAA,EACA,eAAA;AAAA,EACA,aAAA;AAAA,EACA,eAAA;AAAA,EACA,QAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAeR,WAAA,CAAY;AAAA,IACV,WAAA;AAAA,IACA,cAAA;AAAA,IACA,OAAA;AAAA,IACA,aAAA;AAAA,IACA,eAAA;AAAA,IACA;AAAA,GACF,GAOI,EAAC,EAAG;AAKN,IAAA,MAAM,MAAA,GAAS,QAAQ,GAAA,CAAI,eAAA;AAC3B,IAAA,MAAM,YAAA,GAAe,WAAA,EAAa,MAAA,IAAU,cAAA,EAAgB,MAAA,IAAU,MAAA;AACtE,IAAA,MAAM,eAAA,GAAkB,cAAA,EAAgB,MAAA,IAAU,WAAA,EAAa,MAAA,IAAU,MAAA;AAEzE,IAAA,IAAI,CAAC,YAAA,IAAgB,CAAC,eAAA,EAAiB;AACrC,MAAA,MAAM,IAAI,KAAA;AAAA,QACR;AAAA,OACF;AAAA,IACF;AAEA,IAAA,KAAA,CAAM;AAAA,MACJ,WAAA,EAAa;AAAA,QACX,IAAA,EAAM,aAAa,IAAA,IAAQ,eAAA;AAAA,QAC3B,MAAA,EAAQ;AAAA,OACV;AAAA,MACA,cAAA,EAAgB;AAAA,QACd,IAAA,EAAM,gBAAgB,IAAA,IAAQ,uBAAA;AAAA,QAC9B,MAAA,EAAQ;AAAA,OACV;AAAA,MACA,SAAS,OAAA,IAAW;AAAA,KACrB,CAAA;AAED,IAAA,IAAA,CAAK,YAAA,GAAe,YAAA;AACpB,IAAA,IAAA,CAAK,eAAA,GAAkB,eAAA;AACvB,IAAA,IAAA,CAAK,gBAAgB,aAAA,IAAiB,KAAA;AACtC,IAAA,IAAA,CAAK,kBAAkB,eAAA,IAAmB,IAAA;AAC1C,IAAA,IAAA,CAAK,WAAW,QAAA,IAAY,OAAA;AAAA,EAC9B;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,MAAM,WAAA,GAAc;AAClB,IAAA,MAAM,QAAA,GAAW,MAAM,KAAA,CAAM,CAAA,EAAG,gBAAgB,CAAA,iBAAA,CAAA,EAAqB;AAAA,MACnE,SAAS,EAAE,aAAA,EAAe,CAAA,MAAA,EAAS,IAAA,CAAK,YAAY,CAAA,CAAA;AAAG,KACxD,CAAA;AAED,IAAA,IAAI,CAAC,SAAS,EAAA,EAAI;AAChB,MAAA,MAAM,IAAA,GAAO,MAAM,QAAA,CAAS,IAAA,EAAK;AACjC,MAAA,MAAM,IAAI,KAAA,CAAM,CAAA,4BAAA,EAA+B,SAAS,MAAM,CAAA,GAAA,EAAM,IAAI,CAAA,CAAE,CAAA;AAAA,IAC5E;AAEA,IAAA,MAAM,IAAA,GAAQ,MAAM,QAAA,CAAS,IAAA,EAAK;AAWlC,IAAA,OACE,IAAA,CAAK,MAAA,EAAQ,GAAA,CAAI,CAAA,CAAA,MAAM;AAAA,MACrB,SAAS,CAAA,CAAE,OAAA;AAAA,MACX,IAAA,EAAM,CAAA,CAAE,WAAA,IAAe,CAAA,CAAE,OAAA;AAAA,MACzB,QAAA,EAAU,EAAE,QAAA,IAAY,IAAA;AAAA,MACxB,WAAA,EAAa,EAAE,WAAA,IAAe,EAAA;AAAA,MAC9B,IAAA,EAAM,CAAA,CAAE,IAAA,IAAQ,EAAC;AAAA,MACjB,MAAA,EAAQ,EAAE,MAAA,IAAU;AAAA,KACtB,CAAE,KAAK,EAAC;AAAA,EAEZ;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,WAAA,GAAc;AAClB,IAAA,OAAO,EAAE,OAAA,EAAS,CAAC,CAAC,KAAK,cAAA,EAAe;AAAA,EAC1C;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAgBA,MAAM,KAAA,CACJ,KAAA,EACA,OAAA,EACgC;AAChC,IAAA,MAAM,IAAA,GAAO,OAAO,KAAA,KAAU,QAAA,GAAW,QAAQ,MAAM,IAAA,CAAK,eAAe,KAAK,CAAA;AAEhF,IAAA,IAAI,IAAA,CAAK,IAAA,EAAK,CAAE,MAAA,KAAW,CAAA,EAAG;AAC5B,MAAA,MAAM,IAAI,MAAM,qBAAqB,CAAA;AAAA,IACvC;AAEA,IAAA,MAAM,OAAA,GAAU,OAAA,EAAS,OAAA,IAAW,IAAA,CAAK,OAAA;AAEzC,IAAA,MAAM,IAAA,GAAO;AAAA,MACX,IAAA;AAAA,MACA,OAAA,EAAS,OAAA;AAAA,MACT,OAAA,EAAS,IAAA,CAAK,WAAA,EAAa,IAAA,IAAQ,eAAA;AAAA,MACnC,WAAA,EAAa;AAAA,QACX,aAAA,EAAe,OAAA,EAAS,aAAA,IAAiB,IAAA,CAAK,aAAA;AAAA,QAC9C,eAAA,EAAiB,OAAA,EAAS,eAAA,IAAmB,IAAA,CAAK,eAAA;AAAA,QAClD,GAAI,OAAA,EAAS,YAAA,KAAiB,UAAa,EAAE,YAAA,EAAc,QAAQ,YAAA;AAAa,OAClF;AAAA,MACA,GAAI,OAAA,EAAS,WAAA,KAAgB,UAAa,EAAE,WAAA,EAAa,QAAQ,WAAA,EAAY;AAAA,MAC7E,GAAI,OAAA,EAAS,YAAA,KAAiB,UAAa,EAAE,YAAA,EAAc,QAAQ,YAAA,EAAa;AAAA,MAChF,GAAI,OAAA,EAAS,QAAA,KAAa,UAAa,EAAE,QAAA,EAAU,QAAQ,QAAA;AAAS,KACtE;AAEA,IAAA,MAAM,QAAA,GAAW,MAAM,KAAA,CAAM,CAAA,EAAG,gBAAgB,CAAA,oBAAA,CAAA,EAAwB;AAAA,MACtE,MAAA,EAAQ,MAAA;AAAA,MACR,OAAA,EAAS;AAAA,QACP,cAAA,EAAgB,kBAAA;AAAA,QAChB,aAAA,EAAe,CAAA,MAAA,EAAS,IAAA,CAAK,YAAY,CAAA;AAAA,OAC3C;AAAA,MACA,IAAA,EAAM,IAAA,CAAK,SAAA,CAAU,IAAI;AAAA,KAC1B,CAAA;AAED,IAAA,IAAI,CAAC,SAAS,EAAA,EAAI;AAChB,MAAA,MAAM,SAAA,GAAY,MAAM,QAAA,CAAS,IAAA,EAAK;AACtC,MAAA,MAAM,IAAI,KAAA,CAAM,CAAA,oBAAA,EAAuB,SAAS,MAAM,CAAA,GAAA,EAAM,SAAS,CAAA,CAAE,CAAA;AAAA,IACzE;AAEA,IAAA,IAAI,CAAC,SAAS,IAAA,EAAM;AAClB,MAAA,MAAM,IAAI,MAAM,4CAA4C,CAAA;AAAA,IAC9D;AAIA,IAAA,MAAM,YAAA,GAAe,IAAIC,kBAAA,EAAY;AACrC,IAAA,MAAM,MAAA,GAAS,QAAA,CAAS,IAAA,CAAK,SAAA,EAAU;AACvC,IAAA,MAAM,OAAA,GAAU,IAAI,WAAA,EAAY;AAChC,IAAA,IAAI,MAAA,GAAS,EAAA;AAGb,IAAA,MAAM,UAAA,GAAa,OAAO,IAAA,KAAiB;AACzC,MAAA,IAAI,aAAa,SAAA,EAAW;AAC5B,MAAA,IAAI,CAAC,YAAA,CAAa,KAAA,CAAM,IAAI,CAAA,EAAG;AAC7B,QAAA,MAAM,IAAI,OAAA,CAAc,CAAC,OAAA,EAAS,MAAA,KAAW;AAC3C,UAAA,YAAA,CAAa,IAAA,CAAK,SAAS,OAAO,CAAA;AAClC,UAAA,YAAA,CAAa,IAAA,CAAK,SAAS,OAAO,CAAA;AAClC,UAAA,YAAA,CAAa,IAAA,CAAK,SAAS,MAAM,CAAA;AAAA,QACnC,CAAC,CAAA;AAAA,MACH;AAAA,IACF,CAAA;AAGA,IAAA,YAAA,CAAa,IAAA,CAAK,SAAS,MAAM;AAC/B,MAAA,KAAK,MAAA,CAAO,MAAA,EAAO,CAAE,KAAA,CAAM,MAAM;AAAA,MAAC,CAAC,CAAA;AAAA,IACrC,CAAC,CAAA;AAED,IAAA,KAAA,CAAM,YAAY;AAChB,MAAA,IAAI;AACF,QAAA,OAAO,IAAA,EAAM;AACX,UAAA,IAAI,aAAa,SAAA,EAAW;AAC1B,YAAA,KAAK,MAAA,CAAO,MAAA,EAAO,CAAE,KAAA,CAAM,MAAM;AAAA,YAAC,CAAC,CAAA;AACnC,YAAA;AAAA,UACF;AAEA,UAAA,MAAM,EAAE,IAAA,EAAM,KAAA,EAAM,GAAI,MAAM,OAAO,IAAA,EAAK;AAC1C,UAAA,IAAI,IAAA,EAAM;AAER,YAAA,MAAM,SAAA,GAAY,OAAO,IAAA,EAAK;AAC9B,YAAA,IAAI,SAAA,EAAW;AACb,cAAA,MAAM,KAAA,GAAQ,IAAA,CAAK,KAAA,CAAM,SAAS,CAAA;AAClC,cAAA,MAAM,YAAA,GAAe,KAAA,CAAM,MAAA,EAAQ,YAAA,IAAgB,KAAA,CAAM,YAAA;AACzD,cAAA,IAAI,YAAA,EAAc;AAChB,gBAAA,MAAM,UAAA,CAAW,MAAA,CAAO,IAAA,CAAK,YAAA,EAAc,QAAQ,CAAC,CAAA;AAAA,cACtD;AAAA,YACF;AACA,YAAA,YAAA,CAAa,GAAA,EAAI;AACjB,YAAA;AAAA,UACF;AAEA,UAAA,MAAA,IAAU,QAAQ,MAAA,CAAO,KAAA,EAAO,EAAE,MAAA,EAAQ,MAAM,CAAA;AAChD,UAAA,MAAM,KAAA,GAAQ,MAAA,CAAO,KAAA,CAAM,IAAI,CAAA;AAC/B,UAAA,MAAA,GAAS,KAAA,CAAM,KAAI,IAAK,EAAA;AAExB,UAAA,KAAA,MAAW,QAAQ,KAAA,EAAO;AACxB,YAAA,MAAM,OAAA,GAAU,KAAK,IAAA,EAAK;AAC1B,YAAA,IAAI,CAAC,OAAA,EAAS;AAEd,YAAA,IAAI,aAAa,SAAA,EAAW;AAC1B,cAAA,KAAK,MAAA,CAAO,MAAA,EAAO,CAAE,KAAA,CAAM,MAAM;AAAA,cAAC,CAAC,CAAA;AACnC,cAAA;AAAA,YACF;AAEA,YAAA,MAAM,KAAA,GAAQ,IAAA,CAAK,KAAA,CAAM,OAAO,CAAA;AAChC,YAAA,MAAM,YAAA,GAAe,KAAA,CAAM,MAAA,EAAQ,YAAA,IAAgB,KAAA,CAAM,YAAA;AACzD,YAAA,IAAI,YAAA,EAAc;AAChB,cAAA,MAAM,UAAA,CAAW,MAAA,CAAO,IAAA,CAAK,YAAA,EAAc,QAAQ,CAAC,CAAA;AAAA,YACtD;AAAA,UACF;AAAA,QACF;AAAA,MACF,SAAS,GAAA,EAAK;AACZ,QAAA,IAAI,CAAC,aAAa,SAAA,EAAW;AAC3B,UAAA,YAAA,CAAa,OAAA,CAAQ,eAAe,KAAA,GAAQ,GAAA,GAAM,IAAI,KAAA,CAAM,MAAA,CAAO,GAAG,CAAC,CAAC,CAAA;AAAA,QAC1E;AACA,QAAA,KAAK,MAAA,CAAO,MAAA,EAAO,CAAE,KAAA,CAAM,MAAM;AAAA,QAAC,CAAC,CAAA;AAAA,MACrC;AAAA,IACF,CAAA,GAAG,CAAE,KAAA,CAAM,MAAM;AAAA,IAAC,CAAC,CAAA;AAEnB,IAAA,OAAO,YAAA;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EASA,MAAM,MAAA,CAAO,KAAA,EAA8B,OAAA,EAAiD;AAC1F,IAAA,MAAM,SAAmB,EAAC;AAC1B,IAAA,WAAA,MAAiB,SAAS,KAAA,EAAO;AAC/B,MAAA,MAAA,CAAO,IAAA,CAAK,OAAO,KAAA,KAAU,QAAA,GAAW,OAAO,IAAA,CAAK,KAAK,IAAK,KAAgB,CAAA;AAAA,IAChF;AACA,IAAA,MAAM,cAAc,MAAA,CAAO,MAAA,CAAO,MAAM,CAAA,CAAE,SAAS,QAAQ,CAAA;AAE3D,IAAA,MAAM,IAAA,GAAO;AAAA,MACX,gBAAA,EAAkB;AAAA,QAChB,OAAA,EAAS,IAAA,CAAK,cAAA,EAAgB,IAAA,IAAQ,uBAAA;AAAA,QACtC,aAAA,EAAe,SAAS,aAAA,IAAiB,aAAA;AAAA,QACzC,QAAA,EAAU,OAAA,EAAS,QAAA,IAAY,IAAA,CAAK,QAAA;AAAA,QACpC,eAAA,EAAiB,SAAS,eAAA,IAAmB,IAAA;AAAA,QAC7C,gBAAA,EAAkB,SAAS,gBAAA,IAAoB;AAAA,OACjD;AAAA,MACA,SAAA,EAAW;AAAA,QACT,OAAA,EAAS;AAAA;AACX,KACF;AAEA,IAAA,MAAM,QAAA,GAAW,MAAM,KAAA,CAAM,CAAA,EAAG,gBAAgB,CAAA,kBAAA,CAAA,EAAsB;AAAA,MACpE,MAAA,EAAQ,MAAA;AAAA,MACR,OAAA,EAAS;AAAA,QACP,cAAA,EAAgB,kBAAA;AAAA,QAChB,aAAA,EAAe,CAAA,MAAA,EAAS,IAAA,CAAK,eAAe,CAAA;AAAA,OAC9C;AAAA,MACA,IAAA,EAAM,IAAA,CAAK,SAAA,CAAU,IAAI;AAAA,KAC1B,CAAA;AAED,IAAA,IAAI,CAAC,SAAS,EAAA,EAAI;AAChB,MAAA,MAAM,IAAA,GAAO,MAAM,QAAA,CAAS,IAAA,EAAK;AACjC,MAAA,MAAM,IAAI,KAAA,CAAM,CAAA,oBAAA,EAAuB,SAAS,MAAM,CAAA,GAAA,EAAM,IAAI,CAAA,CAAE,CAAA;AAAA,IACpE;AAEA,IAAA,MAAM,MAAA,GAAU,MAAM,QAAA,CAAS,IAAA,EAAK;AAOpC,IAAA,OAAO,MAAA,CAAO,eAAe,UAAA,IAAc,EAAA;AAAA,EAC7C;AAAA,EAEA,MAAc,eAAe,MAAA,EAAgD;AAC3E,IAAA,MAAM,SAAmB,EAAC;AAC1B,IAAA,WAAA,MAAiB,SAAS,MAAA,EAAQ;AAChC,MAAA,MAAA,CAAO,IAAA,CAAK,OAAO,KAAA,KAAU,QAAA,GAAW,OAAO,IAAA,CAAK,KAAK,IAAK,KAAgB,CAAA;AAAA,IAChF;AACA,IAAA,OAAO,MAAA,CAAO,MAAA,CAAO,MAAM,CAAA,CAAE,SAAS,OAAO,CAAA;AAAA,EAC/C;AACF","file":"index.cjs","sourcesContent":["import { PassThrough } from 'node:stream';\nimport { MastraVoice } from '@mastra/core/voice';\n\nconst INWORLD_API_BASE = 'https://api.inworld.ai';\n\ntype InworldTtsModel = 'inworld-tts-2' | 'inworld-tts-1.5-max' | 'inworld-tts-1.5-mini';\n\ntype DeliveryMode = 'STABLE' | 'BALANCED' | 'CREATIVE';\n\ntype InworldSttModel = 'groq/whisper-large-v3';\n\ntype AudioEncoding = 'LINEAR16' | 'MP3' | 'OGG_OPUS' | 'ALAW' | 'MULAW' | 'FLAC' | 'PCM' | 'WAV';\n\ntype SttAudioEncoding = 'LINEAR16' | 'MP3' | 'OGG_OPUS' | 'FLAC' | 'AUTO_DETECT';\n\ninterface InworldVoiceConfig {\n  name?: InworldTtsModel;\n  apiKey?: string;\n}\n\ninterface InworldListeningConfig {\n  name?: InworldSttModel;\n  apiKey?: string;\n}\n\ninterface InworldSpeakOptions {\n  speaker?: string;\n  audioEncoding?: AudioEncoding;\n  sampleRateHertz?: number;\n  speakingRate?: number;\n  temperature?: number;\n  deliveryMode?: DeliveryMode;\n  language?: string;\n}\n\ninterface InworldListenOptions {\n  audioEncoding?: SttAudioEncoding;\n  sampleRateHertz?: number;\n  language?: string;\n  numberOfChannels?: number;\n}\n\nexport type {\n  InworldVoiceConfig,\n  InworldListeningConfig,\n  InworldSpeakOptions,\n  InworldListenOptions,\n  InworldTtsModel,\n  InworldSttModel,\n  AudioEncoding,\n  SttAudioEncoding,\n  DeliveryMode,\n};\n\nexport class InworldVoice extends MastraVoice {\n  private speechApiKey: string;\n  private listeningApiKey: string;\n  private audioEncoding: AudioEncoding;\n  private sampleRateHertz: number;\n  private language: string;\n\n  /**\n   * Creates an instance of the InworldVoice class.\n   *\n   * @param {Object} options - The options for the voice configuration.\n   * @param {InworldVoiceConfig} [options.speechModel] - TTS model config. Default: inworld-tts-2.\n   * @param {InworldListeningConfig} [options.listeningModel] - STT model config. Default: groq/whisper-large-v3.\n   * @param {string} [options.speaker] - Default voice ID. Default: 'Dennis'.\n   * @param {AudioEncoding} [options.audioEncoding] - TTS audio format. Default: 'MP3'.\n   * @param {number} [options.sampleRateHertz] - TTS sample rate. Default: 48000.\n   * @param {string} [options.language] - STT language (BCP-47). Default: 'en-US'.\n   *\n   * @throws {Error} If no API key is provided or found in INWORLD_API_KEY env var.\n   */\n  constructor({\n    speechModel,\n    listeningModel,\n    speaker,\n    audioEncoding,\n    sampleRateHertz,\n    language,\n  }: {\n    speechModel?: InworldVoiceConfig;\n    listeningModel?: InworldListeningConfig;\n    speaker?: string;\n    audioEncoding?: AudioEncoding;\n    sampleRateHertz?: number;\n    language?: string;\n  } = {}) {\n    // Resolve TTS and STT keys independently so callers can scope credentials\n    // per service. If only one model is configured with a key, it is reused as\n    // the fallback for the other (preserving the \"one key works everywhere\"\n    // ergonomic). The INWORLD_API_KEY env var is the final fallback.\n    const envKey = process.env.INWORLD_API_KEY;\n    const speechApiKey = speechModel?.apiKey ?? listeningModel?.apiKey ?? envKey;\n    const listeningApiKey = listeningModel?.apiKey ?? speechModel?.apiKey ?? envKey;\n\n    if (!speechApiKey || !listeningApiKey) {\n      throw new Error(\n        'Inworld API key is required. Pass apiKey in speechModel/listeningModel config or set INWORLD_API_KEY env var.',\n      );\n    }\n\n    super({\n      speechModel: {\n        name: speechModel?.name ?? 'inworld-tts-2',\n        apiKey: speechApiKey,\n      },\n      listeningModel: {\n        name: listeningModel?.name ?? 'groq/whisper-large-v3',\n        apiKey: listeningApiKey,\n      },\n      speaker: speaker ?? 'Dennis',\n    });\n\n    this.speechApiKey = speechApiKey;\n    this.listeningApiKey = listeningApiKey;\n    this.audioEncoding = audioEncoding ?? 'MP3';\n    this.sampleRateHertz = sampleRateHertz ?? 48000;\n    this.language = language ?? 'en-US';\n  }\n\n  /**\n   * Retrieves a list of available voices from the Inworld API.\n   *\n   * @returns {Promise<Array<{ voiceId: string; name: string; language: string; description: string; tags: string[]; source: string }>>}\n   */\n  async getSpeakers() {\n    const response = await fetch(`${INWORLD_API_BASE}/voices/v1/voices`, {\n      headers: { Authorization: `Basic ${this.speechApiKey}` },\n    });\n\n    if (!response.ok) {\n      const text = await response.text();\n      throw new Error(`Inworld list voices failed (${response.status}): ${text}`);\n    }\n\n    const data = (await response.json()) as {\n      voices?: Array<{\n        voiceId: string;\n        displayName?: string;\n        langCode?: string;\n        description?: string;\n        tags?: string[];\n        source?: string;\n      }>;\n    };\n\n    return (\n      data.voices?.map(v => ({\n        voiceId: v.voiceId,\n        name: v.displayName ?? v.voiceId,\n        language: v.langCode ?? 'en',\n        description: v.description ?? '',\n        tags: v.tags ?? [],\n        source: v.source ?? 'SYSTEM',\n      })) ?? []\n    );\n  }\n\n  /**\n   * Checks if listening capabilities are enabled.\n   */\n  async getListener() {\n    return { enabled: !!this.listeningModel };\n  }\n\n  /**\n   * Converts text to speech using Inworld's streaming TTS endpoint.\n   *\n   * Returns a ReadableStream that emits audio chunks progressively as they\n   * arrive from the API — following the same streaming pattern used by the\n   * Deepgram and PlayAI Mastra voice providers.\n   *\n   * Uses the `/tts/v1/voice:stream` endpoint which returns newline-delimited\n   * JSON, each line containing a base64-encoded audio chunk.\n   *\n   * @param {string | NodeJS.ReadableStream} input - Text or stream to convert to speech.\n   * @param {InworldSpeakOptions} [options] - TTS options.\n   * @returns {Promise<NodeJS.ReadableStream>} Progressive audio stream.\n   */\n  async speak(\n    input: string | NodeJS.ReadableStream,\n    options?: InworldSpeakOptions & { speaker?: string },\n  ): Promise<NodeJS.ReadableStream> {\n    const text = typeof input === 'string' ? input : await this.streamToString(input);\n\n    if (text.trim().length === 0) {\n      throw new Error('Input text is empty');\n    }\n\n    const speaker = options?.speaker ?? this.speaker;\n\n    const body = {\n      text,\n      voiceId: speaker,\n      modelId: this.speechModel?.name ?? 'inworld-tts-2',\n      audioConfig: {\n        audioEncoding: options?.audioEncoding ?? this.audioEncoding,\n        sampleRateHertz: options?.sampleRateHertz ?? this.sampleRateHertz,\n        ...(options?.speakingRate !== undefined && { speakingRate: options.speakingRate }),\n      },\n      ...(options?.temperature !== undefined && { temperature: options.temperature }),\n      ...(options?.deliveryMode !== undefined && { deliveryMode: options.deliveryMode }),\n      ...(options?.language !== undefined && { language: options.language }),\n    };\n\n    const response = await fetch(`${INWORLD_API_BASE}/tts/v1/voice:stream`, {\n      method: 'POST',\n      headers: {\n        'Content-Type': 'application/json',\n        Authorization: `Basic ${this.speechApiKey}`,\n      },\n      body: JSON.stringify(body),\n    });\n\n    if (!response.ok) {\n      const errorText = await response.text();\n      throw new Error(`Inworld TTS failed (${response.status}): ${errorText}`);\n    }\n\n    if (!response.body) {\n      throw new Error('Inworld TTS streaming response has no body');\n    }\n\n    // Progressive streaming: return the PassThrough immediately while reading\n    // NDJSON chunks from the response body in a background async task.\n    const outputStream = new PassThrough();\n    const reader = response.body.getReader();\n    const decoder = new TextDecoder();\n    let buffer = '';\n\n    // Write with backpressure: await drain when the consumer is slow.\n    const writeChunk = async (data: Buffer) => {\n      if (outputStream.destroyed) return;\n      if (!outputStream.write(data)) {\n        await new Promise<void>((resolve, reject) => {\n          outputStream.once('drain', resolve);\n          outputStream.once('close', resolve);\n          outputStream.once('error', reject);\n        });\n      }\n    };\n\n    // Cancel the upstream reader when the consumer closes early.\n    outputStream.once('close', () => {\n      void reader.cancel().catch(() => {});\n    });\n\n    void (async () => {\n      try {\n        while (true) {\n          if (outputStream.destroyed) {\n            void reader.cancel().catch(() => {});\n            break;\n          }\n\n          const { done, value } = await reader.read();\n          if (done) {\n            // Process any remaining data in the buffer before ending\n            const remaining = buffer.trim();\n            if (remaining) {\n              const chunk = JSON.parse(remaining);\n              const audioContent = chunk.result?.audioContent ?? chunk.audioContent;\n              if (audioContent) {\n                await writeChunk(Buffer.from(audioContent, 'base64'));\n              }\n            }\n            outputStream.end();\n            break;\n          }\n\n          buffer += decoder.decode(value, { stream: true });\n          const lines = buffer.split('\\n');\n          buffer = lines.pop() ?? '';\n\n          for (const line of lines) {\n            const trimmed = line.trim();\n            if (!trimmed) continue;\n\n            if (outputStream.destroyed) {\n              void reader.cancel().catch(() => {});\n              return;\n            }\n\n            const chunk = JSON.parse(trimmed);\n            const audioContent = chunk.result?.audioContent ?? chunk.audioContent;\n            if (audioContent) {\n              await writeChunk(Buffer.from(audioContent, 'base64'));\n            }\n          }\n        }\n      } catch (err) {\n        if (!outputStream.destroyed) {\n          outputStream.destroy(err instanceof Error ? err : new Error(String(err)));\n        }\n        void reader.cancel().catch(() => {});\n      }\n    })().catch(() => {});\n\n    return outputStream;\n  }\n\n  /**\n   * Converts audio to text using Inworld's batch STT endpoint.\n   *\n   * @param {NodeJS.ReadableStream} input - Audio stream to transcribe.\n   * @param {InworldListenOptions} [options] - STT options.\n   * @returns {Promise<string>} Transcribed text.\n   */\n  async listen(input: NodeJS.ReadableStream, options?: InworldListenOptions): Promise<string> {\n    const chunks: Buffer[] = [];\n    for await (const chunk of input) {\n      chunks.push(typeof chunk === 'string' ? Buffer.from(chunk) : (chunk as Buffer));\n    }\n    const audioBase64 = Buffer.concat(chunks).toString('base64');\n\n    const body = {\n      transcribeConfig: {\n        modelId: this.listeningModel?.name ?? 'groq/whisper-large-v3',\n        audioEncoding: options?.audioEncoding ?? 'AUTO_DETECT',\n        language: options?.language ?? this.language,\n        sampleRateHertz: options?.sampleRateHertz ?? 16000,\n        numberOfChannels: options?.numberOfChannels ?? 1,\n      },\n      audioData: {\n        content: audioBase64,\n      },\n    };\n\n    const response = await fetch(`${INWORLD_API_BASE}/stt/v1/transcribe`, {\n      method: 'POST',\n      headers: {\n        'Content-Type': 'application/json',\n        Authorization: `Basic ${this.listeningApiKey}`,\n      },\n      body: JSON.stringify(body),\n    });\n\n    if (!response.ok) {\n      const text = await response.text();\n      throw new Error(`Inworld STT failed (${response.status}): ${text}`);\n    }\n\n    const result = (await response.json()) as {\n      transcription?: {\n        transcript?: string;\n        isFinal?: boolean;\n      };\n    };\n\n    return result.transcription?.transcript ?? '';\n  }\n\n  private async streamToString(stream: NodeJS.ReadableStream): Promise<string> {\n    const chunks: Buffer[] = [];\n    for await (const chunk of stream) {\n      chunks.push(typeof chunk === 'string' ? Buffer.from(chunk) : (chunk as Buffer));\n    }\n    return Buffer.concat(chunks).toString('utf-8');\n  }\n}\n"]}

package/dist/index.d.ts

@@ -0,0 +1,104 @@
+import { MastraVoice } from '@mastra/core/voice';
+type InworldTtsModel = 'inworld-tts-2' | 'inworld-tts-1.5-max' | 'inworld-tts-1.5-mini';
+type DeliveryMode = 'STABLE' | 'BALANCED' | 'CREATIVE';
+type InworldSttModel = 'groq/whisper-large-v3';
+type AudioEncoding = 'LINEAR16' | 'MP3' | 'OGG_OPUS' | 'ALAW' | 'MULAW' | 'FLAC' | 'PCM' | 'WAV';
+type SttAudioEncoding = 'LINEAR16' | 'MP3' | 'OGG_OPUS' | 'FLAC' | 'AUTO_DETECT';
+interface InworldVoiceConfig {
+    name?: InworldTtsModel;
+    apiKey?: string;
+}
+interface InworldListeningConfig {
+    name?: InworldSttModel;
+    apiKey?: string;
+}
+interface InworldSpeakOptions {
+    speaker?: string;
+    audioEncoding?: AudioEncoding;
+    sampleRateHertz?: number;
+    speakingRate?: number;
+    temperature?: number;
+    deliveryMode?: DeliveryMode;
+    language?: string;
+}
+interface InworldListenOptions {
+    audioEncoding?: SttAudioEncoding;
+    sampleRateHertz?: number;
+    language?: string;
+    numberOfChannels?: number;
+}
+export type { InworldVoiceConfig, InworldListeningConfig, InworldSpeakOptions, InworldListenOptions, InworldTtsModel, InworldSttModel, AudioEncoding, SttAudioEncoding, DeliveryMode, };
+export declare class InworldVoice extends MastraVoice {
+    private speechApiKey;
+    private listeningApiKey;
+    private audioEncoding;
+    private sampleRateHertz;
+    private language;
+    /**
+     * Creates an instance of the InworldVoice class.
+     *
+     * @param {Object} options - The options for the voice configuration.
+     * @param {InworldVoiceConfig} [options.speechModel] - TTS model config. Default: inworld-tts-2.
+     * @param {InworldListeningConfig} [options.listeningModel] - STT model config. Default: groq/whisper-large-v3.
+     * @param {string} [options.speaker] - Default voice ID. Default: 'Dennis'.
+     * @param {AudioEncoding} [options.audioEncoding] - TTS audio format. Default: 'MP3'.
+     * @param {number} [options.sampleRateHertz] - TTS sample rate. Default: 48000.
+     * @param {string} [options.language] - STT language (BCP-47). Default: 'en-US'.
+     *
+     * @throws {Error} If no API key is provided or found in INWORLD_API_KEY env var.
+     */
+    constructor({ speechModel, listeningModel, speaker, audioEncoding, sampleRateHertz, language, }?: {
+        speechModel?: InworldVoiceConfig;
+        listeningModel?: InworldListeningConfig;
+        speaker?: string;
+        audioEncoding?: AudioEncoding;
+        sampleRateHertz?: number;
+        language?: string;
+    });
+    /**
+     * Retrieves a list of available voices from the Inworld API.
+     *
+     * @returns {Promise<Array<{ voiceId: string; name: string; language: string; description: string; tags: string[]; source: string }>>}
+     */
+    getSpeakers(): Promise<{
+        voiceId: string;
+        name: string;
+        language: string;
+        description: string;
+        tags: string[];
+        source: string;
+    }[]>;
+    /**
+     * Checks if listening capabilities are enabled.
+     */
+    getListener(): Promise<{
+        enabled: boolean;
+    }>;
+    /**
+     * Converts text to speech using Inworld's streaming TTS endpoint.
+     *
+     * Returns a ReadableStream that emits audio chunks progressively as they
+     * arrive from the API — following the same streaming pattern used by the
+     * Deepgram and PlayAI Mastra voice providers.
+     *
+     * Uses the `/tts/v1/voice:stream` endpoint which returns newline-delimited
+     * JSON, each line containing a base64-encoded audio chunk.
+     *
+     * @param {string | NodeJS.ReadableStream} input - Text or stream to convert to speech.
+     * @param {InworldSpeakOptions} [options] - TTS options.
+     * @returns {Promise<NodeJS.ReadableStream>} Progressive audio stream.
+     */
+    speak(input: string | NodeJS.ReadableStream, options?: InworldSpeakOptions & {
+        speaker?: string;
+    }): Promise<NodeJS.ReadableStream>;
+    /**
+     * Converts audio to text using Inworld's batch STT endpoint.
+     *
+     * @param {NodeJS.ReadableStream} input - Audio stream to transcribe.
+     * @param {InworldListenOptions} [options] - STT options.
+     * @returns {Promise<string>} Transcribed text.
+     */
+    listen(input: NodeJS.ReadableStream, options?: InworldListenOptions): Promise<string>;
+    private streamToString;
+}
+//# sourceMappingURL=index.d.ts.map

package/dist/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AAIjD,KAAK,eAAe,GAAG,eAAe,GAAG,qBAAqB,GAAG,sBAAsB,CAAC;AAExF,KAAK,YAAY,GAAG,QAAQ,GAAG,UAAU,GAAG,UAAU,CAAC;AAEvD,KAAK,eAAe,GAAG,uBAAuB,CAAC;AAE/C,KAAK,aAAa,GAAG,UAAU,GAAG,KAAK,GAAG,UAAU,GAAG,MAAM,GAAG,OAAO,GAAG,MAAM,GAAG,KAAK,GAAG,KAAK,CAAC;AAEjG,KAAK,gBAAgB,GAAG,UAAU,GAAG,KAAK,GAAG,UAAU,GAAG,MAAM,GAAG,aAAa,CAAC;AAEjF,UAAU,kBAAkB;IAC1B,IAAI,CAAC,EAAE,eAAe,CAAC;IACvB,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AAED,UAAU,sBAAsB;IAC9B,IAAI,CAAC,EAAE,eAAe,CAAC;IACvB,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AAED,UAAU,mBAAmB;IAC3B,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,aAAa,CAAC,EAAE,aAAa,CAAC;IAC9B,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,YAAY,CAAC,EAAE,YAAY,CAAC;IAC5B,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,UAAU,oBAAoB;IAC5B,aAAa,CAAC,EAAE,gBAAgB,CAAC;IACjC,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,gBAAgB,CAAC,EAAE,MAAM,CAAC;CAC3B;AAED,YAAY,EACV,kBAAkB,EAClB,sBAAsB,EACtB,mBAAmB,EACnB,oBAAoB,EACpB,eAAe,EACf,eAAe,EACf,aAAa,EACb,gBAAgB,EAChB,YAAY,GACb,CAAC;AAEF,qBAAa,YAAa,SAAQ,WAAW;IAC3C,OAAO,CAAC,YAAY,CAAS;IAC7B,OAAO,CAAC,eAAe,CAAS;IAChC,OAAO,CAAC,aAAa,CAAgB;IACrC,OAAO,CAAC,eAAe,CAAS;IAChC,OAAO,CAAC,QAAQ,CAAS;IAEzB;;;;;;;;;;;;OAYG;gBACS,EACV,WAAW,EACX,cAAc,EACd,OAAO,EACP,aAAa,EACb,eAAe,EACf,QAAQ,GACT,GAAE;QACD,WAAW,CAAC,EAAE,kBAAkB,CAAC;QACjC,cAAc,CAAC,EAAE,sBAAsB,CAAC;QACxC,OAAO,CAAC,EAAE,MAAM,CAAC;QACjB,aAAa,CAAC,EAAE,aAAa,CAAC;QAC9B,eAAe,CAAC,EAAE,MAAM,CAAC;QACzB,QAAQ,CAAC,EAAE,MAAM,CAAC;KACd;IAkCN;;;;OAIG;IACG,WAAW;;;;;;;;IAiCjB;;OAEG;IACG,WAAW;;;IAIjB;;;;;;;;;;;;;OAaG;IACG,KAAK,CACT,KAAK,EAAE,MAAM,GAAG,MAAM,CAAC,cAAc,EACrC,OAAO,CAAC,EAAE,mBAAmB,GAAG;QAAE,OAAO,CAAC,EAAE,MAAM,CAAA;KAAE,GACnD,OAAO,CAAC,MAAM,CAAC,cAAc,CAAC;IAuHjC;;;;;;OAMG;IACG,MAAM,CAAC,KAAK,EAAE,MAAM,CAAC,cAAc,EAAE,OAAO,CAAC,EAAE,oBAAoB,GAAG,OAAO,CAAC,MAAM,CAAC;YA4C7E,cAAc;CAO7B"}

package/dist/index.js

@@ -0,0 +1,254 @@
+import { PassThrough } from 'stream';
+import { MastraVoice } from '@mastra/core/voice';
+
+// src/index.ts
+var INWORLD_API_BASE = "https://api.inworld.ai";
+var InworldVoice = class extends MastraVoice {
+  speechApiKey;
+  listeningApiKey;
+  audioEncoding;
+  sampleRateHertz;
+  language;
+  /**
+   * Creates an instance of the InworldVoice class.
+   *
+   * @param {Object} options - The options for the voice configuration.
+   * @param {InworldVoiceConfig} [options.speechModel] - TTS model config. Default: inworld-tts-2.
+   * @param {InworldListeningConfig} [options.listeningModel] - STT model config. Default: groq/whisper-large-v3.
+   * @param {string} [options.speaker] - Default voice ID. Default: 'Dennis'.
+   * @param {AudioEncoding} [options.audioEncoding] - TTS audio format. Default: 'MP3'.
+   * @param {number} [options.sampleRateHertz] - TTS sample rate. Default: 48000.
+   * @param {string} [options.language] - STT language (BCP-47). Default: 'en-US'.
+   *
+   * @throws {Error} If no API key is provided or found in INWORLD_API_KEY env var.
+   */
+  constructor({
+    speechModel,
+    listeningModel,
+    speaker,
+    audioEncoding,
+    sampleRateHertz,
+    language
+  } = {}) {
+    const envKey = process.env.INWORLD_API_KEY;
+    const speechApiKey = speechModel?.apiKey ?? listeningModel?.apiKey ?? envKey;
+    const listeningApiKey = listeningModel?.apiKey ?? speechModel?.apiKey ?? envKey;
+    if (!speechApiKey || !listeningApiKey) {
+      throw new Error(
+        "Inworld API key is required. Pass apiKey in speechModel/listeningModel config or set INWORLD_API_KEY env var."
+      );
+    }
+    super({
+      speechModel: {
+        name: speechModel?.name ?? "inworld-tts-2",
+        apiKey: speechApiKey
+      },
+      listeningModel: {
+        name: listeningModel?.name ?? "groq/whisper-large-v3",
+        apiKey: listeningApiKey
+      },
+      speaker: speaker ?? "Dennis"
+    });
+    this.speechApiKey = speechApiKey;
+    this.listeningApiKey = listeningApiKey;
+    this.audioEncoding = audioEncoding ?? "MP3";
+    this.sampleRateHertz = sampleRateHertz ?? 48e3;
+    this.language = language ?? "en-US";
+  }
+  /**
+   * Retrieves a list of available voices from the Inworld API.
+   *
+   * @returns {Promise<Array<{ voiceId: string; name: string; language: string; description: string; tags: string[]; source: string }>>}
+   */
+  async getSpeakers() {
+    const response = await fetch(`${INWORLD_API_BASE}/voices/v1/voices`, {
+      headers: { Authorization: `Basic ${this.speechApiKey}` }
+    });
+    if (!response.ok) {
+      const text = await response.text();
+      throw new Error(`Inworld list voices failed (${response.status}): ${text}`);
+    }
+    const data = await response.json();
+    return data.voices?.map((v) => ({
+      voiceId: v.voiceId,
+      name: v.displayName ?? v.voiceId,
+      language: v.langCode ?? "en",
+      description: v.description ?? "",
+      tags: v.tags ?? [],
+      source: v.source ?? "SYSTEM"
+    })) ?? [];
+  }
+  /**
+   * Checks if listening capabilities are enabled.
+   */
+  async getListener() {
+    return { enabled: !!this.listeningModel };
+  }
+  /**
+   * Converts text to speech using Inworld's streaming TTS endpoint.
+   *
+   * Returns a ReadableStream that emits audio chunks progressively as they
+   * arrive from the API — following the same streaming pattern used by the
+   * Deepgram and PlayAI Mastra voice providers.
+   *
+   * Uses the `/tts/v1/voice:stream` endpoint which returns newline-delimited
+   * JSON, each line containing a base64-encoded audio chunk.
+   *
+   * @param {string | NodeJS.ReadableStream} input - Text or stream to convert to speech.
+   * @param {InworldSpeakOptions} [options] - TTS options.
+   * @returns {Promise<NodeJS.ReadableStream>} Progressive audio stream.
+   */
+  async speak(input, options) {
+    const text = typeof input === "string" ? input : await this.streamToString(input);
+    if (text.trim().length === 0) {
+      throw new Error("Input text is empty");
+    }
+    const speaker = options?.speaker ?? this.speaker;
+    const body = {
+      text,
+      voiceId: speaker,
+      modelId: this.speechModel?.name ?? "inworld-tts-2",
+      audioConfig: {
+        audioEncoding: options?.audioEncoding ?? this.audioEncoding,
+        sampleRateHertz: options?.sampleRateHertz ?? this.sampleRateHertz,
+        ...options?.speakingRate !== void 0 && { speakingRate: options.speakingRate }
+      },
+      ...options?.temperature !== void 0 && { temperature: options.temperature },
+      ...options?.deliveryMode !== void 0 && { deliveryMode: options.deliveryMode },
+      ...options?.language !== void 0 && { language: options.language }
+    };
+    const response = await fetch(`${INWORLD_API_BASE}/tts/v1/voice:stream`, {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+        Authorization: `Basic ${this.speechApiKey}`
+      },
+      body: JSON.stringify(body)
+    });
+    if (!response.ok) {
+      const errorText = await response.text();
+      throw new Error(`Inworld TTS failed (${response.status}): ${errorText}`);
+    }
+    if (!response.body) {
+      throw new Error("Inworld TTS streaming response has no body");
+    }
+    const outputStream = new PassThrough();
+    const reader = response.body.getReader();
+    const decoder = new TextDecoder();
+    let buffer = "";
+    const writeChunk = async (data) => {
+      if (outputStream.destroyed) return;
+      if (!outputStream.write(data)) {
+        await new Promise((resolve, reject) => {
+          outputStream.once("drain", resolve);
+          outputStream.once("close", resolve);
+          outputStream.once("error", reject);
+        });
+      }
+    };
+    outputStream.once("close", () => {
+      void reader.cancel().catch(() => {
+      });
+    });
+    void (async () => {
+      try {
+        while (true) {
+          if (outputStream.destroyed) {
+            void reader.cancel().catch(() => {
+            });
+            break;
+          }
+          const { done, value } = await reader.read();
+          if (done) {
+            const remaining = buffer.trim();
+            if (remaining) {
+              const chunk = JSON.parse(remaining);
+              const audioContent = chunk.result?.audioContent ?? chunk.audioContent;
+              if (audioContent) {
+                await writeChunk(Buffer.from(audioContent, "base64"));
+              }
+            }
+            outputStream.end();
+            break;
+          }
+          buffer += decoder.decode(value, { stream: true });
+          const lines = buffer.split("\n");
+          buffer = lines.pop() ?? "";
+          for (const line of lines) {
+            const trimmed = line.trim();
+            if (!trimmed) continue;
+            if (outputStream.destroyed) {
+              void reader.cancel().catch(() => {
+              });
+              return;
+            }
+            const chunk = JSON.parse(trimmed);
+            const audioContent = chunk.result?.audioContent ?? chunk.audioContent;
+            if (audioContent) {
+              await writeChunk(Buffer.from(audioContent, "base64"));
+            }
+          }
+        }
+      } catch (err) {
+        if (!outputStream.destroyed) {
+          outputStream.destroy(err instanceof Error ? err : new Error(String(err)));
+        }
+        void reader.cancel().catch(() => {
+        });
+      }
+    })().catch(() => {
+    });
+    return outputStream;
+  }
+  /**
+   * Converts audio to text using Inworld's batch STT endpoint.
+   *
+   * @param {NodeJS.ReadableStream} input - Audio stream to transcribe.
+   * @param {InworldListenOptions} [options] - STT options.
+   * @returns {Promise<string>} Transcribed text.
+   */
+  async listen(input, options) {
+    const chunks = [];
+    for await (const chunk of input) {
+      chunks.push(typeof chunk === "string" ? Buffer.from(chunk) : chunk);
+    }
+    const audioBase64 = Buffer.concat(chunks).toString("base64");
+    const body = {
+      transcribeConfig: {
+        modelId: this.listeningModel?.name ?? "groq/whisper-large-v3",
+        audioEncoding: options?.audioEncoding ?? "AUTO_DETECT",
+        language: options?.language ?? this.language,
+        sampleRateHertz: options?.sampleRateHertz ?? 16e3,
+        numberOfChannels: options?.numberOfChannels ?? 1
+      },
+      audioData: {
+        content: audioBase64
+      }
+    };
+    const response = await fetch(`${INWORLD_API_BASE}/stt/v1/transcribe`, {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+        Authorization: `Basic ${this.listeningApiKey}`
+      },
+      body: JSON.stringify(body)
+    });
+    if (!response.ok) {
+      const text = await response.text();
+      throw new Error(`Inworld STT failed (${response.status}): ${text}`);
+    }
+    const result = await response.json();
+    return result.transcription?.transcript ?? "";
+  }
+  async streamToString(stream) {
+    const chunks = [];
+    for await (const chunk of stream) {
+      chunks.push(typeof chunk === "string" ? Buffer.from(chunk) : chunk);
+    }
+    return Buffer.concat(chunks).toString("utf-8");
+  }
+};
+
+export { InworldVoice };
+//# sourceMappingURL=index.js.map
+//# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/index.ts"],"names":[],"mappings":";;;;AAGA,IAAM,gBAAA,GAAmB,wBAAA;AAmDlB,IAAM,YAAA,GAAN,cAA2B,WAAA,CAAY;AAAA,EACpC,YAAA;AAAA,EACA,eAAA;AAAA,EACA,aAAA;AAAA,EACA,eAAA;AAAA,EACA,QAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAeR,WAAA,CAAY;AAAA,IACV,WAAA;AAAA,IACA,cAAA;AAAA,IACA,OAAA;AAAA,IACA,aAAA;AAAA,IACA,eAAA;AAAA,IACA;AAAA,GACF,GAOI,EAAC,EAAG;AAKN,IAAA,MAAM,MAAA,GAAS,QAAQ,GAAA,CAAI,eAAA;AAC3B,IAAA,MAAM,YAAA,GAAe,WAAA,EAAa,MAAA,IAAU,cAAA,EAAgB,MAAA,IAAU,MAAA;AACtE,IAAA,MAAM,eAAA,GAAkB,cAAA,EAAgB,MAAA,IAAU,WAAA,EAAa,MAAA,IAAU,MAAA;AAEzE,IAAA,IAAI,CAAC,YAAA,IAAgB,CAAC,eAAA,EAAiB;AACrC,MAAA,MAAM,IAAI,KAAA;AAAA,QACR;AAAA,OACF;AAAA,IACF;AAEA,IAAA,KAAA,CAAM;AAAA,MACJ,WAAA,EAAa;AAAA,QACX,IAAA,EAAM,aAAa,IAAA,IAAQ,eAAA;AAAA,QAC3B,MAAA,EAAQ;AAAA,OACV;AAAA,MACA,cAAA,EAAgB;AAAA,QACd,IAAA,EAAM,gBAAgB,IAAA,IAAQ,uBAAA;AAAA,QAC9B,MAAA,EAAQ;AAAA,OACV;AAAA,MACA,SAAS,OAAA,IAAW;AAAA,KACrB,CAAA;AAED,IAAA,IAAA,CAAK,YAAA,GAAe,YAAA;AACpB,IAAA,IAAA,CAAK,eAAA,GAAkB,eAAA;AACvB,IAAA,IAAA,CAAK,gBAAgB,aAAA,IAAiB,KAAA;AACtC,IAAA,IAAA,CAAK,kBAAkB,eAAA,IAAmB,IAAA;AAC1C,IAAA,IAAA,CAAK,WAAW,QAAA,IAAY,OAAA;AAAA,EAC9B;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,MAAM,WAAA,GAAc;AAClB,IAAA,MAAM,QAAA,GAAW,MAAM,KAAA,CAAM,CAAA,EAAG,gBAAgB,CAAA,iBAAA,CAAA,EAAqB;AAAA,MACnE,SAAS,EAAE,aAAA,EAAe,CAAA,MAAA,EAAS,IAAA,CAAK,YAAY,CAAA,CAAA;AAAG,KACxD,CAAA;AAED,IAAA,IAAI,CAAC,SAAS,EAAA,EAAI;AAChB,MAAA,MAAM,IAAA,GAAO,MAAM,QAAA,CAAS,IAAA,EAAK;AACjC,MAAA,MAAM,IAAI,KAAA,CAAM,CAAA,4BAAA,EAA+B,SAAS,MAAM,CAAA,GAAA,EAAM,IAAI,CAAA,CAAE,CAAA;AAAA,IAC5E;AAEA,IAAA,MAAM,IAAA,GAAQ,MAAM,QAAA,CAAS,IAAA,EAAK;AAWlC,IAAA,OACE,IAAA,CAAK,MAAA,EAAQ,GAAA,CAAI,CAAA,CAAA,MAAM;AAAA,MACrB,SAAS,CAAA,CAAE,OAAA;AAAA,MACX,IAAA,EAAM,CAAA,CAAE,WAAA,IAAe,CAAA,CAAE,OAAA;AAAA,MACzB,QAAA,EAAU,EAAE,QAAA,IAAY,IAAA;AAAA,MACxB,WAAA,EAAa,EAAE,WAAA,IAAe,EAAA;AAAA,MAC9B,IAAA,EAAM,CAAA,CAAE,IAAA,IAAQ,EAAC;AAAA,MACjB,MAAA,EAAQ,EAAE,MAAA,IAAU;AAAA,KACtB,CAAE,KAAK,EAAC;AAAA,EAEZ;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,WAAA,GAAc;AAClB,IAAA,OAAO,EAAE,OAAA,EAAS,CAAC,CAAC,KAAK,cAAA,EAAe;AAAA,EAC1C;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAgBA,MAAM,KAAA,CACJ,KAAA,EACA,OAAA,EACgC;AAChC,IAAA,MAAM,IAAA,GAAO,OAAO,KAAA,KAAU,QAAA,GAAW,QAAQ,MAAM,IAAA,CAAK,eAAe,KAAK,CAAA;AAEhF,IAAA,IAAI,IAAA,CAAK,IAAA,EAAK,CAAE,MAAA,KAAW,CAAA,EAAG;AAC5B,MAAA,MAAM,IAAI,MAAM,qBAAqB,CAAA;AAAA,IACvC;AAEA,IAAA,MAAM,OAAA,GAAU,OAAA,EAAS,OAAA,IAAW,IAAA,CAAK,OAAA;AAEzC,IAAA,MAAM,IAAA,GAAO;AAAA,MACX,IAAA;AAAA,MACA,OAAA,EAAS,OAAA;AAAA,MACT,OAAA,EAAS,IAAA,CAAK,WAAA,EAAa,IAAA,IAAQ,eAAA;AAAA,MACnC,WAAA,EAAa;AAAA,QACX,aAAA,EAAe,OAAA,EAAS,aAAA,IAAiB,IAAA,CAAK,aAAA;AAAA,QAC9C,eAAA,EAAiB,OAAA,EAAS,eAAA,IAAmB,IAAA,CAAK,eAAA;AAAA,QAClD,GAAI,OAAA,EAAS,YAAA,KAAiB,UAAa,EAAE,YAAA,EAAc,QAAQ,YAAA;AAAa,OAClF;AAAA,MACA,GAAI,OAAA,EAAS,WAAA,KAAgB,UAAa,EAAE,WAAA,EAAa,QAAQ,WAAA,EAAY;AAAA,MAC7E,GAAI,OAAA,EAAS,YAAA,KAAiB,UAAa,EAAE,YAAA,EAAc,QAAQ,YAAA,EAAa;AAAA,MAChF,GAAI,OAAA,EAAS,QAAA,KAAa,UAAa,EAAE,QAAA,EAAU,QAAQ,QAAA;AAAS,KACtE;AAEA,IAAA,MAAM,QAAA,GAAW,MAAM,KAAA,CAAM,CAAA,EAAG,gBAAgB,CAAA,oBAAA,CAAA,EAAwB;AAAA,MACtE,MAAA,EAAQ,MAAA;AAAA,MACR,OAAA,EAAS;AAAA,QACP,cAAA,EAAgB,kBAAA;AAAA,QAChB,aAAA,EAAe,CAAA,MAAA,EAAS,IAAA,CAAK,YAAY,CAAA;AAAA,OAC3C;AAAA,MACA,IAAA,EAAM,IAAA,CAAK,SAAA,CAAU,IAAI;AAAA,KAC1B,CAAA;AAED,IAAA,IAAI,CAAC,SAAS,EAAA,EAAI;AAChB,MAAA,MAAM,SAAA,GAAY,MAAM,QAAA,CAAS,IAAA,EAAK;AACtC,MAAA,MAAM,IAAI,KAAA,CAAM,CAAA,oBAAA,EAAuB,SAAS,MAAM,CAAA,GAAA,EAAM,SAAS,CAAA,CAAE,CAAA;AAAA,IACzE;AAEA,IAAA,IAAI,CAAC,SAAS,IAAA,EAAM;AAClB,MAAA,MAAM,IAAI,MAAM,4CAA4C,CAAA;AAAA,IAC9D;AAIA,IAAA,MAAM,YAAA,GAAe,IAAI,WAAA,EAAY;AACrC,IAAA,MAAM,MAAA,GAAS,QAAA,CAAS,IAAA,CAAK,SAAA,EAAU;AACvC,IAAA,MAAM,OAAA,GAAU,IAAI,WAAA,EAAY;AAChC,IAAA,IAAI,MAAA,GAAS,EAAA;AAGb,IAAA,MAAM,UAAA,GAAa,OAAO,IAAA,KAAiB;AACzC,MAAA,IAAI,aAAa,SAAA,EAAW;AAC5B,MAAA,IAAI,CAAC,YAAA,CAAa,KAAA,CAAM,IAAI,CAAA,EAAG;AAC7B,QAAA,MAAM,IAAI,OAAA,CAAc,CAAC,OAAA,EAAS,MAAA,KAAW;AAC3C,UAAA,YAAA,CAAa,IAAA,CAAK,SAAS,OAAO,CAAA;AAClC,UAAA,YAAA,CAAa,IAAA,CAAK,SAAS,OAAO,CAAA;AAClC,UAAA,YAAA,CAAa,IAAA,CAAK,SAAS,MAAM,CAAA;AAAA,QACnC,CAAC,CAAA;AAAA,MACH;AAAA,IACF,CAAA;AAGA,IAAA,YAAA,CAAa,IAAA,CAAK,SAAS,MAAM;AAC/B,MAAA,KAAK,MAAA,CAAO,MAAA,EAAO,CAAE,KAAA,CAAM,MAAM;AAAA,MAAC,CAAC,CAAA;AAAA,IACrC,CAAC,CAAA;AAED,IAAA,KAAA,CAAM,YAAY;AAChB,MAAA,IAAI;AACF,QAAA,OAAO,IAAA,EAAM;AACX,UAAA,IAAI,aAAa,SAAA,EAAW;AAC1B,YAAA,KAAK,MAAA,CAAO,MAAA,EAAO,CAAE,KAAA,CAAM,MAAM;AAAA,YAAC,CAAC,CAAA;AACnC,YAAA;AAAA,UACF;AAEA,UAAA,MAAM,EAAE,IAAA,EAAM,KAAA,EAAM,GAAI,MAAM,OAAO,IAAA,EAAK;AAC1C,UAAA,IAAI,IAAA,EAAM;AAER,YAAA,MAAM,SAAA,GAAY,OAAO,IAAA,EAAK;AAC9B,YAAA,IAAI,SAAA,EAAW;AACb,cAAA,MAAM,KAAA,GAAQ,IAAA,CAAK,KAAA,CAAM,SAAS,CAAA;AAClC,cAAA,MAAM,YAAA,GAAe,KAAA,CAAM,MAAA,EAAQ,YAAA,IAAgB,KAAA,CAAM,YAAA;AACzD,cAAA,IAAI,YAAA,EAAc;AAChB,gBAAA,MAAM,UAAA,CAAW,MAAA,CAAO,IAAA,CAAK,YAAA,EAAc,QAAQ,CAAC,CAAA;AAAA,cACtD;AAAA,YACF;AACA,YAAA,YAAA,CAAa,GAAA,EAAI;AACjB,YAAA;AAAA,UACF;AAEA,UAAA,MAAA,IAAU,QAAQ,MAAA,CAAO,KAAA,EAAO,EAAE,MAAA,EAAQ,MAAM,CAAA;AAChD,UAAA,MAAM,KAAA,GAAQ,MAAA,CAAO,KAAA,CAAM,IAAI,CAAA;AAC/B,UAAA,MAAA,GAAS,KAAA,CAAM,KAAI,IAAK,EAAA;AAExB,UAAA,KAAA,MAAW,QAAQ,KAAA,EAAO;AACxB,YAAA,MAAM,OAAA,GAAU,KAAK,IAAA,EAAK;AAC1B,YAAA,IAAI,CAAC,OAAA,EAAS;AAEd,YAAA,IAAI,aAAa,SAAA,EAAW;AAC1B,cAAA,KAAK,MAAA,CAAO,MAAA,EAAO,CAAE,KAAA,CAAM,MAAM;AAAA,cAAC,CAAC,CAAA;AACnC,cAAA;AAAA,YACF;AAEA,YAAA,MAAM,KAAA,GAAQ,IAAA,CAAK,KAAA,CAAM,OAAO,CAAA;AAChC,YAAA,MAAM,YAAA,GAAe,KAAA,CAAM,MAAA,EAAQ,YAAA,IAAgB,KAAA,CAAM,YAAA;AACzD,YAAA,IAAI,YAAA,EAAc;AAChB,cAAA,MAAM,UAAA,CAAW,MAAA,CAAO,IAAA,CAAK,YAAA,EAAc,QAAQ,CAAC,CAAA;AAAA,YACtD;AAAA,UACF;AAAA,QACF;AAAA,MACF,SAAS,GAAA,EAAK;AACZ,QAAA,IAAI,CAAC,aAAa,SAAA,EAAW;AAC3B,UAAA,YAAA,CAAa,OAAA,CAAQ,eAAe,KAAA,GAAQ,GAAA,GAAM,IAAI,KAAA,CAAM,MAAA,CAAO,GAAG,CAAC,CAAC,CAAA;AAAA,QAC1E;AACA,QAAA,KAAK,MAAA,CAAO,MAAA,EAAO,CAAE,KAAA,CAAM,MAAM;AAAA,QAAC,CAAC,CAAA;AAAA,MACrC;AAAA,IACF,CAAA,GAAG,CAAE,KAAA,CAAM,MAAM;AAAA,IAAC,CAAC,CAAA;AAEnB,IAAA,OAAO,YAAA;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EASA,MAAM,MAAA,CAAO,KAAA,EAA8B,OAAA,EAAiD;AAC1F,IAAA,MAAM,SAAmB,EAAC;AAC1B,IAAA,WAAA,MAAiB,SAAS,KAAA,EAAO;AAC/B,MAAA,MAAA,CAAO,IAAA,CAAK,OAAO,KAAA,KAAU,QAAA,GAAW,OAAO,IAAA,CAAK,KAAK,IAAK,KAAgB,CAAA;AAAA,IAChF;AACA,IAAA,MAAM,cAAc,MAAA,CAAO,MAAA,CAAO,MAAM,CAAA,CAAE,SAAS,QAAQ,CAAA;AAE3D,IAAA,MAAM,IAAA,GAAO;AAAA,MACX,gBAAA,EAAkB;AAAA,QAChB,OAAA,EAAS,IAAA,CAAK,cAAA,EAAgB,IAAA,IAAQ,uBAAA;AAAA,QACtC,aAAA,EAAe,SAAS,aAAA,IAAiB,aAAA;AAAA,QACzC,QAAA,EAAU,OAAA,EAAS,QAAA,IAAY,IAAA,CAAK,QAAA;AAAA,QACpC,eAAA,EAAiB,SAAS,eAAA,IAAmB,IAAA;AAAA,QAC7C,gBAAA,EAAkB,SAAS,gBAAA,IAAoB;AAAA,OACjD;AAAA,MACA,SAAA,EAAW;AAAA,QACT,OAAA,EAAS;AAAA;AACX,KACF;AAEA,IAAA,MAAM,QAAA,GAAW,MAAM,KAAA,CAAM,CAAA,EAAG,gBAAgB,CAAA,kBAAA,CAAA,EAAsB;AAAA,MACpE,MAAA,EAAQ,MAAA;AAAA,MACR,OAAA,EAAS;AAAA,QACP,cAAA,EAAgB,kBAAA;AAAA,QAChB,aAAA,EAAe,CAAA,MAAA,EAAS,IAAA,CAAK,eAAe,CAAA;AAAA,OAC9C;AAAA,MACA,IAAA,EAAM,IAAA,CAAK,SAAA,CAAU,IAAI;AAAA,KAC1B,CAAA;AAED,IAAA,IAAI,CAAC,SAAS,EAAA,EAAI;AAChB,MAAA,MAAM,IAAA,GAAO,MAAM,QAAA,CAAS,IAAA,EAAK;AACjC,MAAA,MAAM,IAAI,KAAA,CAAM,CAAA,oBAAA,EAAuB,SAAS,MAAM,CAAA,GAAA,EAAM,IAAI,CAAA,CAAE,CAAA;AAAA,IACpE;AAEA,IAAA,MAAM,MAAA,GAAU,MAAM,QAAA,CAAS,IAAA,EAAK;AAOpC,IAAA,OAAO,MAAA,CAAO,eAAe,UAAA,IAAc,EAAA;AAAA,EAC7C;AAAA,EAEA,MAAc,eAAe,MAAA,EAAgD;AAC3E,IAAA,MAAM,SAAmB,EAAC;AAC1B,IAAA,WAAA,MAAiB,SAAS,MAAA,EAAQ;AAChC,MAAA,MAAA,CAAO,IAAA,CAAK,OAAO,KAAA,KAAU,QAAA,GAAW,OAAO,IAAA,CAAK,KAAK,IAAK,KAAgB,CAAA;AAAA,IAChF;AACA,IAAA,OAAO,MAAA,CAAO,MAAA,CAAO,MAAM,CAAA,CAAE,SAAS,OAAO,CAAA;AAAA,EAC/C;AACF","file":"index.js","sourcesContent":["import { PassThrough } from 'node:stream';\nimport { MastraVoice } from '@mastra/core/voice';\n\nconst INWORLD_API_BASE = 'https://api.inworld.ai';\n\ntype InworldTtsModel = 'inworld-tts-2' | 'inworld-tts-1.5-max' | 'inworld-tts-1.5-mini';\n\ntype DeliveryMode = 'STABLE' | 'BALANCED' | 'CREATIVE';\n\ntype InworldSttModel = 'groq/whisper-large-v3';\n\ntype AudioEncoding = 'LINEAR16' | 'MP3' | 'OGG_OPUS' | 'ALAW' | 'MULAW' | 'FLAC' | 'PCM' | 'WAV';\n\ntype SttAudioEncoding = 'LINEAR16' | 'MP3' | 'OGG_OPUS' | 'FLAC' | 'AUTO_DETECT';\n\ninterface InworldVoiceConfig {\n  name?: InworldTtsModel;\n  apiKey?: string;\n}\n\ninterface InworldListeningConfig {\n  name?: InworldSttModel;\n  apiKey?: string;\n}\n\ninterface InworldSpeakOptions {\n  speaker?: string;\n  audioEncoding?: AudioEncoding;\n  sampleRateHertz?: number;\n  speakingRate?: number;\n  temperature?: number;\n  deliveryMode?: DeliveryMode;\n  language?: string;\n}\n\ninterface InworldListenOptions {\n  audioEncoding?: SttAudioEncoding;\n  sampleRateHertz?: number;\n  language?: string;\n  numberOfChannels?: number;\n}\n\nexport type {\n  InworldVoiceConfig,\n  InworldListeningConfig,\n  InworldSpeakOptions,\n  InworldListenOptions,\n  InworldTtsModel,\n  InworldSttModel,\n  AudioEncoding,\n  SttAudioEncoding,\n  DeliveryMode,\n};\n\nexport class InworldVoice extends MastraVoice {\n  private speechApiKey: string;\n  private listeningApiKey: string;\n  private audioEncoding: AudioEncoding;\n  private sampleRateHertz: number;\n  private language: string;\n\n  /**\n   * Creates an instance of the InworldVoice class.\n   *\n   * @param {Object} options - The options for the voice configuration.\n   * @param {InworldVoiceConfig} [options.speechModel] - TTS model config. Default: inworld-tts-2.\n   * @param {InworldListeningConfig} [options.listeningModel] - STT model config. Default: groq/whisper-large-v3.\n   * @param {string} [options.speaker] - Default voice ID. Default: 'Dennis'.\n   * @param {AudioEncoding} [options.audioEncoding] - TTS audio format. Default: 'MP3'.\n   * @param {number} [options.sampleRateHertz] - TTS sample rate. Default: 48000.\n   * @param {string} [options.language] - STT language (BCP-47). Default: 'en-US'.\n   *\n   * @throws {Error} If no API key is provided or found in INWORLD_API_KEY env var.\n   */\n  constructor({\n    speechModel,\n    listeningModel,\n    speaker,\n    audioEncoding,\n    sampleRateHertz,\n    language,\n  }: {\n    speechModel?: InworldVoiceConfig;\n    listeningModel?: InworldListeningConfig;\n    speaker?: string;\n    audioEncoding?: AudioEncoding;\n    sampleRateHertz?: number;\n    language?: string;\n  } = {}) {\n    // Resolve TTS and STT keys independently so callers can scope credentials\n    // per service. If only one model is configured with a key, it is reused as\n    // the fallback for the other (preserving the \"one key works everywhere\"\n    // ergonomic). The INWORLD_API_KEY env var is the final fallback.\n    const envKey = process.env.INWORLD_API_KEY;\n    const speechApiKey = speechModel?.apiKey ?? listeningModel?.apiKey ?? envKey;\n    const listeningApiKey = listeningModel?.apiKey ?? speechModel?.apiKey ?? envKey;\n\n    if (!speechApiKey || !listeningApiKey) {\n      throw new Error(\n        'Inworld API key is required. Pass apiKey in speechModel/listeningModel config or set INWORLD_API_KEY env var.',\n      );\n    }\n\n    super({\n      speechModel: {\n        name: speechModel?.name ?? 'inworld-tts-2',\n        apiKey: speechApiKey,\n      },\n      listeningModel: {\n        name: listeningModel?.name ?? 'groq/whisper-large-v3',\n        apiKey: listeningApiKey,\n      },\n      speaker: speaker ?? 'Dennis',\n    });\n\n    this.speechApiKey = speechApiKey;\n    this.listeningApiKey = listeningApiKey;\n    this.audioEncoding = audioEncoding ?? 'MP3';\n    this.sampleRateHertz = sampleRateHertz ?? 48000;\n    this.language = language ?? 'en-US';\n  }\n\n  /**\n   * Retrieves a list of available voices from the Inworld API.\n   *\n   * @returns {Promise<Array<{ voiceId: string; name: string; language: string; description: string; tags: string[]; source: string }>>}\n   */\n  async getSpeakers() {\n    const response = await fetch(`${INWORLD_API_BASE}/voices/v1/voices`, {\n      headers: { Authorization: `Basic ${this.speechApiKey}` },\n    });\n\n    if (!response.ok) {\n      const text = await response.text();\n      throw new Error(`Inworld list voices failed (${response.status}): ${text}`);\n    }\n\n    const data = (await response.json()) as {\n      voices?: Array<{\n        voiceId: string;\n        displayName?: string;\n        langCode?: string;\n        description?: string;\n        tags?: string[];\n        source?: string;\n      }>;\n    };\n\n    return (\n      data.voices?.map(v => ({\n        voiceId: v.voiceId,\n        name: v.displayName ?? v.voiceId,\n        language: v.langCode ?? 'en',\n        description: v.description ?? '',\n        tags: v.tags ?? [],\n        source: v.source ?? 'SYSTEM',\n      })) ?? []\n    );\n  }\n\n  /**\n   * Checks if listening capabilities are enabled.\n   */\n  async getListener() {\n    return { enabled: !!this.listeningModel };\n  }\n\n  /**\n   * Converts text to speech using Inworld's streaming TTS endpoint.\n   *\n   * Returns a ReadableStream that emits audio chunks progressively as they\n   * arrive from the API — following the same streaming pattern used by the\n   * Deepgram and PlayAI Mastra voice providers.\n   *\n   * Uses the `/tts/v1/voice:stream` endpoint which returns newline-delimited\n   * JSON, each line containing a base64-encoded audio chunk.\n   *\n   * @param {string | NodeJS.ReadableStream} input - Text or stream to convert to speech.\n   * @param {InworldSpeakOptions} [options] - TTS options.\n   * @returns {Promise<NodeJS.ReadableStream>} Progressive audio stream.\n   */\n  async speak(\n    input: string | NodeJS.ReadableStream,\n    options?: InworldSpeakOptions & { speaker?: string },\n  ): Promise<NodeJS.ReadableStream> {\n    const text = typeof input === 'string' ? input : await this.streamToString(input);\n\n    if (text.trim().length === 0) {\n      throw new Error('Input text is empty');\n    }\n\n    const speaker = options?.speaker ?? this.speaker;\n\n    const body = {\n      text,\n      voiceId: speaker,\n      modelId: this.speechModel?.name ?? 'inworld-tts-2',\n      audioConfig: {\n        audioEncoding: options?.audioEncoding ?? this.audioEncoding,\n        sampleRateHertz: options?.sampleRateHertz ?? this.sampleRateHertz,\n        ...(options?.speakingRate !== undefined && { speakingRate: options.speakingRate }),\n      },\n      ...(options?.temperature !== undefined && { temperature: options.temperature }),\n      ...(options?.deliveryMode !== undefined && { deliveryMode: options.deliveryMode }),\n      ...(options?.language !== undefined && { language: options.language }),\n    };\n\n    const response = await fetch(`${INWORLD_API_BASE}/tts/v1/voice:stream`, {\n      method: 'POST',\n      headers: {\n        'Content-Type': 'application/json',\n        Authorization: `Basic ${this.speechApiKey}`,\n      },\n      body: JSON.stringify(body),\n    });\n\n    if (!response.ok) {\n      const errorText = await response.text();\n      throw new Error(`Inworld TTS failed (${response.status}): ${errorText}`);\n    }\n\n    if (!response.body) {\n      throw new Error('Inworld TTS streaming response has no body');\n    }\n\n    // Progressive streaming: return the PassThrough immediately while reading\n    // NDJSON chunks from the response body in a background async task.\n    const outputStream = new PassThrough();\n    const reader = response.body.getReader();\n    const decoder = new TextDecoder();\n    let buffer = '';\n\n    // Write with backpressure: await drain when the consumer is slow.\n    const writeChunk = async (data: Buffer) => {\n      if (outputStream.destroyed) return;\n      if (!outputStream.write(data)) {\n        await new Promise<void>((resolve, reject) => {\n          outputStream.once('drain', resolve);\n          outputStream.once('close', resolve);\n          outputStream.once('error', reject);\n        });\n      }\n    };\n\n    // Cancel the upstream reader when the consumer closes early.\n    outputStream.once('close', () => {\n      void reader.cancel().catch(() => {});\n    });\n\n    void (async () => {\n      try {\n        while (true) {\n          if (outputStream.destroyed) {\n            void reader.cancel().catch(() => {});\n            break;\n          }\n\n          const { done, value } = await reader.read();\n          if (done) {\n            // Process any remaining data in the buffer before ending\n            const remaining = buffer.trim();\n            if (remaining) {\n              const chunk = JSON.parse(remaining);\n              const audioContent = chunk.result?.audioContent ?? chunk.audioContent;\n              if (audioContent) {\n                await writeChunk(Buffer.from(audioContent, 'base64'));\n              }\n            }\n            outputStream.end();\n            break;\n          }\n\n          buffer += decoder.decode(value, { stream: true });\n          const lines = buffer.split('\\n');\n          buffer = lines.pop() ?? '';\n\n          for (const line of lines) {\n            const trimmed = line.trim();\n            if (!trimmed) continue;\n\n            if (outputStream.destroyed) {\n              void reader.cancel().catch(() => {});\n              return;\n            }\n\n            const chunk = JSON.parse(trimmed);\n            const audioContent = chunk.result?.audioContent ?? chunk.audioContent;\n            if (audioContent) {\n              await writeChunk(Buffer.from(audioContent, 'base64'));\n            }\n          }\n        }\n      } catch (err) {\n        if (!outputStream.destroyed) {\n          outputStream.destroy(err instanceof Error ? err : new Error(String(err)));\n        }\n        void reader.cancel().catch(() => {});\n      }\n    })().catch(() => {});\n\n    return outputStream;\n  }\n\n  /**\n   * Converts audio to text using Inworld's batch STT endpoint.\n   *\n   * @param {NodeJS.ReadableStream} input - Audio stream to transcribe.\n   * @param {InworldListenOptions} [options] - STT options.\n   * @returns {Promise<string>} Transcribed text.\n   */\n  async listen(input: NodeJS.ReadableStream, options?: InworldListenOptions): Promise<string> {\n    const chunks: Buffer[] = [];\n    for await (const chunk of input) {\n      chunks.push(typeof chunk === 'string' ? Buffer.from(chunk) : (chunk as Buffer));\n    }\n    const audioBase64 = Buffer.concat(chunks).toString('base64');\n\n    const body = {\n      transcribeConfig: {\n        modelId: this.listeningModel?.name ?? 'groq/whisper-large-v3',\n        audioEncoding: options?.audioEncoding ?? 'AUTO_DETECT',\n        language: options?.language ?? this.language,\n        sampleRateHertz: options?.sampleRateHertz ?? 16000,\n        numberOfChannels: options?.numberOfChannels ?? 1,\n      },\n      audioData: {\n        content: audioBase64,\n      },\n    };\n\n    const response = await fetch(`${INWORLD_API_BASE}/stt/v1/transcribe`, {\n      method: 'POST',\n      headers: {\n        'Content-Type': 'application/json',\n        Authorization: `Basic ${this.listeningApiKey}`,\n      },\n      body: JSON.stringify(body),\n    });\n\n    if (!response.ok) {\n      const text = await response.text();\n      throw new Error(`Inworld STT failed (${response.status}): ${text}`);\n    }\n\n    const result = (await response.json()) as {\n      transcription?: {\n        transcript?: string;\n        isFinal?: boolean;\n      };\n    };\n\n    return result.transcription?.transcript ?? '';\n  }\n\n  private async streamToString(stream: NodeJS.ReadableStream): Promise<string> {\n    const chunks: Buffer[] = [];\n    for await (const chunk of stream) {\n      chunks.push(typeof chunk === 'string' ? Buffer.from(chunk) : (chunk as Buffer));\n    }\n    return Buffer.concat(chunks).toString('utf-8');\n  }\n}\n"]}

package/package.json

@@ -0,0 +1,62 @@
+{
+  "name": "@mastra/voice-inworld",
+  "version": "0.0.0",
+  "description": "Mastra Inworld AI voice integration — streaming TTS and batch STT",
+  "type": "module",
+  "files": [
+    "dist",
+    "CHANGELOG.md"
+  ],
+  "main": "dist/index.js",
+  "types": "dist/index.d.ts",
+  "exports": {
+    ".": {
+      "import": {
+        "types": "./dist/index.d.ts",
+        "default": "./dist/index.js"
+      },
+      "require": {
+        "types": "./dist/index.d.ts",
+        "default": "./dist/index.cjs"
+      }
+    },
+    "./package.json": "./package.json"
+  },
+  "scripts": {
+    "build": "tsup --silent --config tsup.config.ts",
+    "prepack": "pnpx tsx ../../scripts/generate-package-docs.ts",
+    "build:watch": "tsup --watch --silent --config tsup.config.ts",
+    "test": "vitest run",
+    "lint": "eslint ."
+  },
+  "license": "Apache-2.0",
+  "dependencies": {},
+  "devDependencies": {
+    "@internal/lint": "workspace:*",
+    "@internal/types-builder": "workspace:*",
+    "@mastra/core": "workspace:*",
+    "@types/node": "22.19.15",
+    "@vitest/coverage-v8": "catalog:",
+    "@vitest/ui": "catalog:",
+    "eslint": "^10.2.1",
+    "tsup": "^8.5.1",
+    "typescript": "catalog:",
+    "vitest": "catalog:"
+  },
+  "peerDependencies": {
+    "@mastra/core": ">=1.0.0-0 <2.0.0-0",
+    "zod": "^3.25.0 || ^4.0.0"
+  },
+  "homepage": "https://mastra.ai",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/mastra-ai/mastra.git",
+    "directory": "voice/inworld"
+  },
+  "bugs": {
+    "url": "https://github.com/mastra-ai/mastra/issues"
+  },
+  "engines": {
+    "node": ">=22.13.0"
+  }
+}