@zhin.js/speech 0.0.1

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 凉菜
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,38 @@
+# @zhin.js/speech
+
+Zhin.js 可选语音包：STT（Whisper）与 TTS（edge / OpenAI / Azure / 自定义 OpenAI 兼容 API）。
+
+安装：
+
+```bash
+pnpm add @zhin.js/speech
+```
+
+配置（`zhin.config.yml`）：
+
+```yaml
+speech:
+  stt:
+    provider: ollama
+    model: whisper
+    host: http://localhost:11434
+  tts:
+    provider: edge
+    voice: zh-CN-XiaoxiaoNeural
+    edgeTtsCommand: edge-tts
+```
+
+由 `zhin.js` 启动时自动注册 `voice_stt` / `voice_tts` 工具与入站 STT（`ai.multimodal.audio.strategy: transcribe`）。
+
+## TTS Providers
+
+| provider | 说明 |
+|----------|------|
+| `edge`（默认） | 需 `pip install edge-tts` |
+| `openai` | `POST /v1/audio/speech` |
+| `azure` | Azure Cognitive Speech REST |
+| `custom` | OpenAI 兼容 `baseUrl` |
+
+## License
+
+MIT

package/lib/index.d.ts

@@ -0,0 +1,6 @@
+export { createSpeechPipeline } from './pipeline.js';
+export { resolveTtsProvider } from './tts/index.js';
+export type { SpeechConfig, SpeechLogger, SpeechPipeline, STTConfig, TTSConfig, TranscribeInput, TtsProvider, TtsProviderId, TtsSynthesizeInput, TtsSynthesizeResult, SttProviderId, } from './types.js';
+/** 动态 import 时使用的包名（与 package.json name 一致） */
+export declare const SPEECH_PACKAGE = "@zhin.js/speech";
+//# sourceMappingURL=index.d.ts.map

package/lib/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,MAAM,eAAe,CAAC;AACrD,OAAO,EAAE,kBAAkB,EAAE,MAAM,gBAAgB,CAAC;AACpD,YAAY,EACV,YAAY,EACZ,YAAY,EACZ,cAAc,EACd,SAAS,EACT,SAAS,EACT,eAAe,EACf,WAAW,EACX,aAAa,EACb,kBAAkB,EAClB,mBAAmB,EACnB,aAAa,GACd,MAAM,YAAY,CAAC;AAEpB,+CAA+C;AAC/C,eAAO,MAAM,cAAc,oBAAoB,CAAC"}

package/lib/index.js

@@ -0,0 +1,5 @@
+export { createSpeechPipeline } from './pipeline.js';
+export { resolveTtsProvider } from './tts/index.js';
+/** 动态 import 时使用的包名（与 package.json name 一致） */
+export const SPEECH_PACKAGE = '@zhin.js/speech';
+//# sourceMappingURL=index.js.map

package/lib/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,MAAM,eAAe,CAAC;AACrD,OAAO,EAAE,kBAAkB,EAAE,MAAM,gBAAgB,CAAC;AAepD,+CAA+C;AAC/C,MAAM,CAAC,MAAM,cAAc,GAAG,iBAAiB,CAAC"}

package/lib/pipeline.d.ts

@@ -0,0 +1,3 @@
+import type { SpeechConfig, SpeechLogger, SpeechPipeline } from './types.js';
+export declare function createSpeechPipeline(config?: SpeechConfig, _logger?: SpeechLogger): SpeechPipeline;
+//# sourceMappingURL=pipeline.d.ts.map

package/lib/pipeline.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"pipeline.d.ts","sourceRoot":"","sources":["../src/pipeline.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EACV,YAAY,EACZ,YAAY,EACZ,cAAc,EAGf,MAAM,YAAY,CAAC;AA0BpB,wBAAgB,oBAAoB,CAClC,MAAM,CAAC,EAAE,YAAY,EACrB,OAAO,CAAC,EAAE,YAAY,GACrB,cAAc,CA4BhB"}

package/lib/pipeline.js

@@ -0,0 +1,53 @@
+import { transcribeWithOllama } from './stt/ollama.js';
+import { transcribeWithOpenAI } from './stt/openai.js';
+import { resolveTtsProvider } from './tts/index.js';
+const DEFAULT_SPEECH_CONFIG = {
+    stt: {
+        enabled: true,
+        provider: 'ollama',
+        model: 'whisper',
+        host: 'http://localhost:11434',
+    },
+    tts: {
+        enabled: true,
+        provider: 'edge',
+        voice: 'zh-CN-XiaoxiaoNeural',
+        rate: '+0%',
+        pitch: '+0Hz',
+        edgeTtsCommand: 'edge-tts',
+    },
+};
+function mergeSpeechConfig(config) {
+    return {
+        stt: { ...DEFAULT_SPEECH_CONFIG.stt, ...config?.stt },
+        tts: { ...DEFAULT_SPEECH_CONFIG.tts, ...config?.tts },
+    };
+}
+export function createSpeechPipeline(config, _logger) {
+    const merged = mergeSpeechConfig(config);
+    const sttConfig = merged.stt || {};
+    const defaultTts = resolveTtsProvider(merged);
+    return {
+        async transcribe(input) {
+            if (sttConfig.enabled === false) {
+                throw new Error('STT is disabled in speech config');
+            }
+            const provider = sttConfig.provider || 'ollama';
+            const mimeType = input.mimeType || 'audio/wav';
+            if (provider === 'openai') {
+                return transcribeWithOpenAI(input.data, sttConfig, mimeType);
+            }
+            return transcribeWithOllama(input.data, sttConfig, mimeType);
+        },
+        async synthesize(input) {
+            if (merged.tts?.enabled === false) {
+                throw new Error('TTS is disabled in speech config');
+            }
+            const provider = input.provider
+                ? resolveTtsProvider(merged, input.provider)
+                : defaultTts;
+            return provider.synthesize(input);
+        },
+    };
+}
+//# sourceMappingURL=pipeline.js.map

package/lib/pipeline.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pipeline.js","sourceRoot":"","sources":["../src/pipeline.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,MAAM,iBAAiB,CAAC;AACvD,OAAO,EAAE,oBAAoB,EAAE,MAAM,iBAAiB,CAAC;AACvD,OAAO,EAAE,kBAAkB,EAAE,MAAM,gBAAgB,CAAC;AASpD,MAAM,qBAAqB,GAAiB;IAC1C,GAAG,EAAE;QACH,OAAO,EAAE,IAAI;QACb,QAAQ,EAAE,QAAQ;QAClB,KAAK,EAAE,SAAS;QAChB,IAAI,EAAE,wBAAwB;KAC/B;IACD,GAAG,EAAE;QACH,OAAO,EAAE,IAAI;QACb,QAAQ,EAAE,MAAM;QAChB,KAAK,EAAE,sBAAsB;QAC7B,IAAI,EAAE,KAAK;QACX,KAAK,EAAE,MAAM;QACb,cAAc,EAAE,UAAU;KAC3B;CACF,CAAC;AAEF,SAAS,iBAAiB,CAAC,MAAqB;IAC9C,OAAO;QACL,GAAG,EAAE,EAAE,GAAG,qBAAqB,CAAC,GAAG,EAAE,GAAG,MAAM,EAAE,GAAG,EAAE;QACrD,GAAG,EAAE,EAAE,GAAG,qBAAqB,CAAC,GAAG,EAAE,GAAG,MAAM,EAAE,GAAG,EAAE;KACtD,CAAC;AACJ,CAAC;AAED,MAAM,UAAU,oBAAoB,CAClC,MAAqB,EACrB,OAAsB;IAEtB,MAAM,MAAM,GAAG,iBAAiB,CAAC,MAAM,CAAC,CAAC;IACzC,MAAM,SAAS,GAAG,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC;IACnC,MAAM,UAAU,GAAG,kBAAkB,CAAC,MAAM,CAAC,CAAC;IAE9C,OAAO;QACL,KAAK,CAAC,UAAU,CAAC,KAAsB;YACrC,IAAI,SAAS,CAAC,OAAO,KAAK,KAAK,EAAE,CAAC;gBAChC,MAAM,IAAI,KAAK,CAAC,kCAAkC,CAAC,CAAC;YACtD,CAAC;YACD,MAAM,QAAQ,GAAG,SAAS,CAAC,QAAQ,IAAI,QAAQ,CAAC;YAChD,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,IAAI,WAAW,CAAC;YAC/C,IAAI,QAAQ,KAAK,QAAQ,EAAE,CAAC;gBAC1B,OAAO,oBAAoB,CAAC,KAAK,CAAC,IAAI,EAAE,SAAS,EAAE,QAAQ,CAAC,CAAC;YAC/D,CAAC;YACD,OAAO,oBAAoB,CAAC,KAAK,CAAC,IAAI,EAAE,SAAS,EAAE,QAAQ,CAAC,CAAC;QAC/D,CAAC;QAED,KAAK,CAAC,UAAU,CAAC,KAAyB;YACxC,IAAI,MAAM,CAAC,GAAG,EAAE,OAAO,KAAK,KAAK,EAAE,CAAC;gBAClC,MAAM,IAAI,KAAK,CAAC,kCAAkC,CAAC,CAAC;YACtD,CAAC;YACD,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ;gBAC7B,CAAC,CAAC,kBAAkB,CAAC,MAAM,EAAE,KAAK,CAAC,QAAQ,CAAC;gBAC5C,CAAC,CAAC,UAAU,CAAC;YACf,OAAO,QAAQ,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;QACpC,CAAC;KACF,CAAC;AACJ,CAAC"}

package/lib/stt/ollama.d.ts

@@ -0,0 +1,3 @@
+import type { STTConfig } from '../types.js';
+export declare function transcribeWithOllama(audioData: Buffer, sttConfig: STTConfig, _mimeType?: string): Promise<string>;
+//# sourceMappingURL=ollama.d.ts.map

package/lib/stt/ollama.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"ollama.d.ts","sourceRoot":"","sources":["../../src/stt/ollama.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AAE7C,wBAAsB,oBAAoB,CACxC,SAAS,EAAE,MAAM,EACjB,SAAS,EAAE,SAAS,EACpB,SAAS,GAAE,MAAoB,GAC9B,OAAO,CAAC,MAAM,CAAC,CAyBjB"}

package/lib/stt/ollama.js

@@ -0,0 +1,24 @@
+export async function transcribeWithOllama(audioData, sttConfig, _mimeType = 'audio/wav') {
+    const host = sttConfig.host || 'http://localhost:11434';
+    const model = sttConfig.model || 'whisper';
+    const base64Audio = audioData.toString('base64');
+    const response = await fetch(`${host}/api/chat`, {
+        method: 'POST',
+        headers: { 'Content-Type': 'application/json' },
+        body: JSON.stringify({
+            model,
+            messages: [{
+                    role: 'user',
+                    content: '请将这段音频转写为文字，只输出转写的文字内容，不要加任何说明。',
+                    images: [base64Audio],
+                }],
+            stream: false,
+        }),
+    });
+    if (!response.ok) {
+        throw new Error(`Ollama STT failed: ${response.status} ${response.statusText}`);
+    }
+    const data = await response.json();
+    return data.message?.content?.trim() || '';
+}
+//# sourceMappingURL=ollama.js.map

package/lib/stt/ollama.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ollama.js","sourceRoot":"","sources":["../../src/stt/ollama.ts"],"names":[],"mappings":"AAEA,MAAM,CAAC,KAAK,UAAU,oBAAoB,CACxC,SAAiB,EACjB,SAAoB,EACpB,YAAoB,WAAW;IAE/B,MAAM,IAAI,GAAG,SAAS,CAAC,IAAI,IAAI,wBAAwB,CAAC;IACxD,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,IAAI,SAAS,CAAC;IAC3C,MAAM,WAAW,GAAG,SAAS,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC;IAEjD,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,GAAG,IAAI,WAAW,EAAE;QAC/C,MAAM,EAAE,MAAM;QACd,OAAO,EAAE,EAAE,cAAc,EAAE,kBAAkB,EAAE;QAC/C,IAAI,EAAE,IAAI,CAAC,SAAS,CAAC;YACnB,KAAK;YACL,QAAQ,EAAE,CAAC;oBACT,IAAI,EAAE,MAAM;oBACZ,OAAO,EAAE,iCAAiC;oBAC1C,MAAM,EAAE,CAAC,WAAW,CAAC;iBACtB,CAAC;YACF,MAAM,EAAE,KAAK;SACd,CAAC;KACH,CAAC,CAAC;IAEH,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;QACjB,MAAM,IAAI,KAAK,CAAC,sBAAsB,QAAQ,CAAC,MAAM,IAAI,QAAQ,CAAC,UAAU,EAAE,CAAC,CAAC;IAClF,CAAC;IAED,MAAM,IAAI,GAAG,MAAM,QAAQ,CAAC,IAAI,EAAwC,CAAC;IACzE,OAAO,IAAI,CAAC,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;AAC7C,CAAC"}

package/lib/stt/openai.d.ts

@@ -0,0 +1,3 @@
+import type { STTConfig } from '../types.js';
+export declare function transcribeWithOpenAI(audioData: Buffer, sttConfig: STTConfig, mimeType?: string): Promise<string>;
+//# sourceMappingURL=openai.d.ts.map

package/lib/stt/openai.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"openai.d.ts","sourceRoot":"","sources":["../../src/stt/openai.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AAE7C,wBAAsB,oBAAoB,CACxC,SAAS,EAAE,MAAM,EACjB,SAAS,EAAE,SAAS,EACpB,QAAQ,GAAE,MAAoB,GAC7B,OAAO,CAAC,MAAM,CAAC,CA+BjB"}

package/lib/stt/openai.js

@@ -0,0 +1,33 @@
+import { randomBytes } from 'node:crypto';
+import { tmpdir } from 'node:os';
+import { join } from 'node:path';
+import { unlink, writeFile } from 'node:fs/promises';
+export async function transcribeWithOpenAI(audioData, sttConfig, mimeType = 'audio/wav') {
+    const host = sttConfig.host || 'https://api.openai.com';
+    const model = sttConfig.model || 'whisper-1';
+    const apiKey = sttConfig.apiKey || '';
+    const tmpFile = join(tmpdir(), `zhin-stt-${randomBytes(8).toString('hex')}.wav`);
+    await writeFile(tmpFile, audioData);
+    try {
+        const ext = mimeType.includes('mp3') ? 'mp3' : 'wav';
+        const formData = new FormData();
+        const blob = new Blob([new Uint8Array(audioData)], { type: mimeType });
+        formData.append('file', blob, `audio.${ext}`);
+        formData.append('model', model);
+        formData.append('language', 'zh');
+        const response = await fetch(`${host}/v1/audio/transcriptions`, {
+            method: 'POST',
+            headers: { Authorization: `Bearer ${apiKey}` },
+            body: formData,
+        });
+        if (!response.ok) {
+            throw new Error(`OpenAI STT failed: ${response.status} ${response.statusText}`);
+        }
+        const data = await response.json();
+        return data.text?.trim() || '';
+    }
+    finally {
+        await unlink(tmpFile).catch(() => { });
+    }
+}
+//# sourceMappingURL=openai.js.map

package/lib/stt/openai.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"openai.js","sourceRoot":"","sources":["../../src/stt/openai.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AAC1C,OAAO,EAAE,MAAM,EAAE,MAAM,SAAS,CAAC;AACjC,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AACjC,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAC;AAGrD,MAAM,CAAC,KAAK,UAAU,oBAAoB,CACxC,SAAiB,EACjB,SAAoB,EACpB,WAAmB,WAAW;IAE9B,MAAM,IAAI,GAAG,SAAS,CAAC,IAAI,IAAI,wBAAwB,CAAC;IACxD,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,IAAI,WAAW,CAAC;IAC7C,MAAM,MAAM,GAAG,SAAS,CAAC,MAAM,IAAI,EAAE,CAAC;IAEtC,MAAM,OAAO,GAAG,IAAI,CAAC,MAAM,EAAE,EAAE,YAAY,WAAW,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;IACjF,MAAM,SAAS,CAAC,OAAO,EAAE,SAAS,CAAC,CAAC;IAEpC,IAAI,CAAC;QACH,MAAM,GAAG,GAAG,QAAQ,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC;QACrD,MAAM,QAAQ,GAAG,IAAI,QAAQ,EAAE,CAAC;QAChC,MAAM,IAAI,GAAG,IAAI,IAAI,CAAC,CAAC,IAAI,UAAU,CAAC,SAAS,CAAC,CAAC,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,CAAC;QACvE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,IAAI,EAAE,SAAS,GAAG,EAAE,CAAC,CAAC;QAC9C,QAAQ,CAAC,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;QAChC,QAAQ,CAAC,MAAM,CAAC,UAAU,EAAE,IAAI,CAAC,CAAC;QAElC,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,GAAG,IAAI,0BAA0B,EAAE;YAC9D,MAAM,EAAE,MAAM;YACd,OAAO,EAAE,EAAE,aAAa,EAAE,UAAU,MAAM,EAAE,EAAE;YAC9C,IAAI,EAAE,QAAQ;SACf,CAAC,CAAC;QAEH,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;YACjB,MAAM,IAAI,KAAK,CAAC,sBAAsB,QAAQ,CAAC,MAAM,IAAI,QAAQ,CAAC,UAAU,EAAE,CAAC,CAAC;QAClF,CAAC;QAED,MAAM,IAAI,GAAG,MAAM,QAAQ,CAAC,IAAI,EAAuB,CAAC;QACxD,OAAO,IAAI,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;IACjC,CAAC;YAAS,CAAC;QACT,MAAM,MAAM,CAAC,OAAO,CAAC,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAC;IACxC,CAAC;AACH,CAAC"}

package/lib/tts/azure.d.ts

@@ -0,0 +1,4 @@
+import type { TTSConfig, TtsProvider } from '../types.js';
+/** Azure Cognitive Services text-to-speech REST (output format: audio-16khz-128kbitrate-mono-mp3) */
+export declare function createAzureTtsProvider(config: TTSConfig): TtsProvider;
+//# sourceMappingURL=azure.d.ts.map

package/lib/tts/azure.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"azure.d.ts","sourceRoot":"","sources":["../../src/tts/azure.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,SAAS,EAAE,WAAW,EAA2C,MAAM,aAAa,CAAC;AAEnG,qGAAqG;AACrG,wBAAgB,sBAAsB,CAAC,MAAM,EAAE,SAAS,GAAG,WAAW,CA6BrE"}

package/lib/tts/azure.js

@@ -0,0 +1,36 @@
+/** Azure Cognitive Services text-to-speech REST (output format: audio-16khz-128kbitrate-mono-mp3) */
+export function createAzureTtsProvider(config) {
+    return {
+        id: 'azure',
+        async synthesize(input) {
+            const region = config.region || 'eastasia';
+            const subscriptionKey = config.subscriptionKey || '';
+            const voice = input.voice || config.voice || 'zh-CN-XiaoxiaoNeural';
+            const url = `https://${region}.tts.speech.microsoft.com/cognitiveservices/v1`;
+            const ssml = `<speak version='1.0' xml:lang='zh-CN'><voice name='${voice}'>${escapeXml(input.text)}</voice></speak>`;
+            const response = await fetch(url, {
+                method: 'POST',
+                headers: {
+                    'Ocp-Apim-Subscription-Key': subscriptionKey,
+                    'Content-Type': 'application/ssml+xml',
+                    'X-Microsoft-OutputFormat': 'audio-16khz-128kbitrate-mono-mp3',
+                },
+                body: ssml,
+            });
+            if (!response.ok) {
+                throw new Error(`Azure TTS failed: ${response.status} ${response.statusText}`);
+            }
+            const data = Buffer.from(await response.arrayBuffer());
+            return { data, format: 'mp3' };
+        },
+    };
+}
+function escapeXml(text) {
+    return text
+        .replace(/&/g, '&amp;')
+        .replace(/</g, '&lt;')
+        .replace(/>/g, '&gt;')
+        .replace(/"/g, '&quot;')
+        .replace(/'/g, '&apos;');
+}
+//# sourceMappingURL=azure.js.map

package/lib/tts/azure.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"azure.js","sourceRoot":"","sources":["../../src/tts/azure.ts"],"names":[],"mappings":"AAEA,qGAAqG;AACrG,MAAM,UAAU,sBAAsB,CAAC,MAAiB;IACtD,OAAO;QACL,EAAE,EAAE,OAAO;QACX,KAAK,CAAC,UAAU,CAAC,KAAyB;YACxC,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,IAAI,UAAU,CAAC;YAC3C,MAAM,eAAe,GAAG,MAAM,CAAC,eAAe,IAAI,EAAE,CAAC;YACrD,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,IAAI,MAAM,CAAC,KAAK,IAAI,sBAAsB,CAAC;YACpE,MAAM,GAAG,GAAG,WAAW,MAAM,gDAAgD,CAAC;YAE9E,MAAM,IAAI,GAAG,sDAAsD,KAAK,KAAK,SAAS,CAAC,KAAK,CAAC,IAAI,CAAC,kBAAkB,CAAC;YAErH,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,GAAG,EAAE;gBAChC,MAAM,EAAE,MAAM;gBACd,OAAO,EAAE;oBACP,2BAA2B,EAAE,eAAe;oBAC5C,cAAc,EAAE,sBAAsB;oBACtC,0BAA0B,EAAE,kCAAkC;iBAC/D;gBACD,IAAI,EAAE,IAAI;aACX,CAAC,CAAC;YAEH,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;gBACjB,MAAM,IAAI,KAAK,CAAC,qBAAqB,QAAQ,CAAC,MAAM,IAAI,QAAQ,CAAC,UAAU,EAAE,CAAC,CAAC;YACjF,CAAC;YAED,MAAM,IAAI,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,QAAQ,CAAC,WAAW,EAAE,CAAC,CAAC;YACvD,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC;QACjC,CAAC;KACF,CAAC;AACJ,CAAC;AAED,SAAS,SAAS,CAAC,IAAY;IAC7B,OAAO,IAAI;SACR,OAAO,CAAC,IAAI,EAAE,OAAO,CAAC;SACtB,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;SACrB,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;SACrB,OAAO,CAAC,IAAI,EAAE,QAAQ,CAAC;SACvB,OAAO,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;AAC7B,CAAC"}

package/lib/tts/custom.d.ts

@@ -0,0 +1,3 @@
+import type { SpeechConfig, TtsProvider } from '../types.js';
+export declare function createCustomTtsProvider(config: SpeechConfig): TtsProvider;
+//# sourceMappingURL=custom.d.ts.map

package/lib/tts/custom.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"custom.d.ts","sourceRoot":"","sources":["../../src/tts/custom.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAa,WAAW,EAA2C,MAAM,aAAa,CAAC;AAMjH,wBAAgB,uBAAuB,CAAC,MAAM,EAAE,YAAY,GAAG,WAAW,CA2CzE"}

package/lib/tts/custom.js

@@ -0,0 +1,46 @@
+function resolveOpenAiApiKey(config) {
+    return config.tts?.apiKey || config.stt?.apiKey || '';
+}
+export function createCustomTtsProvider(config) {
+    const ttsConfig = config.tts || {};
+    return {
+        id: 'custom',
+        async synthesize(input) {
+            const baseUrl = (ttsConfig.baseUrl || '').replace(/\/$/, '');
+            if (!baseUrl) {
+                throw new Error('custom TTS requires speech.tts.baseUrl');
+            }
+            const model = ttsConfig.model || 'tts-1';
+            const voice = input.voice || ttsConfig.voice || 'default';
+            const apiKey = resolveOpenAiApiKey(config);
+            const format = input.format === 'wav' ? 'wav' : 'mp3';
+            const headers = {
+                'Content-Type': 'application/json',
+                ...(ttsConfig.headers || {}),
+            };
+            if (apiKey)
+                headers.Authorization = `Bearer ${apiKey}`;
+            const body = {
+                model,
+                input: input.text,
+                voice,
+                response_format: format,
+            };
+            if (input.speed != null)
+                body.speed = input.speed;
+            else if (ttsConfig.speed != null)
+                body.speed = ttsConfig.speed;
+            const response = await fetch(baseUrl, {
+                method: 'POST',
+                headers,
+                body: JSON.stringify(body),
+            });
+            if (!response.ok) {
+                throw new Error(`Custom TTS failed: ${response.status} ${response.statusText}`);
+            }
+            const data = Buffer.from(await response.arrayBuffer());
+            return { data, format };
+        },
+    };
+}
+//# sourceMappingURL=custom.js.map

package/lib/tts/custom.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"custom.js","sourceRoot":"","sources":["../../src/tts/custom.ts"],"names":[],"mappings":"AAEA,SAAS,mBAAmB,CAAC,MAAoB;IAC/C,OAAO,MAAM,CAAC,GAAG,EAAE,MAAM,IAAI,MAAM,CAAC,GAAG,EAAE,MAAM,IAAI,EAAE,CAAC;AACxD,CAAC;AAED,MAAM,UAAU,uBAAuB,CAAC,MAAoB;IAC1D,MAAM,SAAS,GAAG,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC;IACnC,OAAO;QACL,EAAE,EAAE,QAAQ;QACZ,KAAK,CAAC,UAAU,CAAC,KAAyB;YACxC,MAAM,OAAO,GAAG,CAAC,SAAS,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;YAC7D,IAAI,CAAC,OAAO,EAAE,CAAC;gBACb,MAAM,IAAI,KAAK,CAAC,wCAAwC,CAAC,CAAC;YAC5D,CAAC;YACD,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,IAAI,OAAO,CAAC;YACzC,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,IAAI,SAAS,CAAC,KAAK,IAAI,SAAS,CAAC;YAC1D,MAAM,MAAM,GAAG,mBAAmB,CAAC,MAAM,CAAC,CAAC;YAC3C,MAAM,MAAM,GAAG,KAAK,CAAC,MAAM,KAAK,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC;YAEtD,MAAM,OAAO,GAA2B;gBACtC,cAAc,EAAE,kBAAkB;gBAClC,GAAG,CAAC,SAAS,CAAC,OAAO,IAAI,EAAE,CAAC;aAC7B,CAAC;YACF,IAAI,MAAM;gBAAE,OAAO,CAAC,aAAa,GAAG,UAAU,MAAM,EAAE,CAAC;YAEvD,MAAM,IAAI,GAA4B;gBACpC,KAAK;gBACL,KAAK,EAAE,KAAK,CAAC,IAAI;gBACjB,KAAK;gBACL,eAAe,EAAE,MAAM;aACxB,CAAC;YACF,IAAI,KAAK,CAAC,KAAK,IAAI,IAAI;gBAAE,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC;iBAC7C,IAAI,SAAS,CAAC,KAAK,IAAI,IAAI;gBAAE,IAAI,CAAC,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC;YAE/D,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,OAAO,EAAE;gBACpC,MAAM,EAAE,MAAM;gBACd,OAAO;gBACP,IAAI,EAAE,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC;aAC3B,CAAC,CAAC;YAEH,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;gBACjB,MAAM,IAAI,KAAK,CAAC,sBAAsB,QAAQ,CAAC,MAAM,IAAI,QAAQ,CAAC,UAAU,EAAE,CAAC,CAAC;YAClF,CAAC;YAED,MAAM,IAAI,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,QAAQ,CAAC,WAAW,EAAE,CAAC,CAAC;YACvD,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC;QAC1B,CAAC;KACF,CAAC;AACJ,CAAC"}

package/lib/tts/edge.d.ts

@@ -0,0 +1,3 @@
+import type { TTSConfig, TtsProvider } from '../types.js';
+export declare function createEdgeTtsProvider(config: TTSConfig): TtsProvider;
+//# sourceMappingURL=edge.d.ts.map

package/lib/tts/edge.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"edge.d.ts","sourceRoot":"","sources":["../../src/tts/edge.ts"],"names":[],"mappings":"AAMA,OAAO,KAAK,EAAE,SAAS,EAAE,WAAW,EAA2C,MAAM,aAAa,CAAC;AAInG,wBAAgB,qBAAqB,CAAC,MAAM,EAAE,SAAS,GAAG,WAAW,CAuBpE"}

package/lib/tts/edge.js

@@ -0,0 +1,29 @@
+import { exec } from 'node:child_process';
+import { randomBytes } from 'node:crypto';
+import { readFile, unlink } from 'node:fs/promises';
+import { tmpdir } from 'node:os';
+import { join } from 'node:path';
+import { promisify } from 'node:util';
+const execAsync = promisify(exec);
+export function createEdgeTtsProvider(config) {
+    return {
+        id: 'edge',
+        async synthesize(input) {
+            const voice = input.voice || config.voice || 'zh-CN-XiaoxiaoNeural';
+            const rate = config.rate || '+0%';
+            const pitch = config.pitch || '+0Hz';
+            const cmd = config.edgeTtsCommand || 'edge-tts';
+            const tmpFile = join(tmpdir(), `zhin-tts-${randomBytes(8).toString('hex')}.mp3`);
+            const safeText = input.text.replace(/"/g, '\\"').replace(/'/g, "\\'");
+            try {
+                await execAsync(`${cmd} --voice "${voice}" --rate="${rate}" --pitch="${pitch}" --text "${safeText}" --write-media "${tmpFile}"`, { timeout: 30_000 });
+                const data = await readFile(tmpFile);
+                return { data, format: 'mp3' };
+            }
+            finally {
+                await unlink(tmpFile).catch(() => { });
+            }
+        },
+    };
+}
+//# sourceMappingURL=edge.js.map

package/lib/tts/edge.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"edge.js","sourceRoot":"","sources":["../../src/tts/edge.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,IAAI,EAAE,MAAM,oBAAoB,CAAC;AAC1C,OAAO,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AAC1C,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,kBAAkB,CAAC;AACpD,OAAO,EAAE,MAAM,EAAE,MAAM,SAAS,CAAC;AACjC,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AACjC,OAAO,EAAE,SAAS,EAAE,MAAM,WAAW,CAAC;AAGtC,MAAM,SAAS,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC;AAElC,MAAM,UAAU,qBAAqB,CAAC,MAAiB;IACrD,OAAO;QACL,EAAE,EAAE,MAAM;QACV,KAAK,CAAC,UAAU,CAAC,KAAyB;YACxC,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,IAAI,MAAM,CAAC,KAAK,IAAI,sBAAsB,CAAC;YACpE,MAAM,IAAI,GAAG,MAAM,CAAC,IAAI,IAAI,KAAK,CAAC;YAClC,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,IAAI,MAAM,CAAC;YACrC,MAAM,GAAG,GAAG,MAAM,CAAC,cAAc,IAAI,UAAU,CAAC;YAChD,MAAM,OAAO,GAAG,IAAI,CAAC,MAAM,EAAE,EAAE,YAAY,WAAW,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;YACjF,MAAM,QAAQ,GAAG,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;YAEtE,IAAI,CAAC;gBACH,MAAM,SAAS,CACb,GAAG,GAAG,aAAa,KAAK,aAAa,IAAI,cAAc,KAAK,aAAa,QAAQ,oBAAoB,OAAO,GAAG,EAC/G,EAAE,OAAO,EAAE,MAAM,EAAE,CACpB,CAAC;gBACF,MAAM,IAAI,GAAG,MAAM,QAAQ,CAAC,OAAO,CAAC,CAAC;gBACrC,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC;YACjC,CAAC;oBAAS,CAAC;gBACT,MAAM,MAAM,CAAC,OAAO,CAAC,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAC;YACxC,CAAC;QACH,CAAC;KACF,CAAC;AACJ,CAAC"}

package/lib/tts/index.d.ts

@@ -0,0 +1,3 @@
+import type { SpeechConfig, TtsProvider, TtsProviderId } from '../types.js';
+export declare function resolveTtsProvider(config: SpeechConfig, override?: TtsProviderId): TtsProvider;
+//# sourceMappingURL=index.d.ts.map

package/lib/tts/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/tts/index.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAE,WAAW,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AAM5E,wBAAgB,kBAAkB,CAAC,MAAM,EAAE,YAAY,EAAE,QAAQ,CAAC,EAAE,aAAa,GAAG,WAAW,CAe9F"}

package/lib/tts/index.js

@@ -0,0 +1,20 @@
+import { createAzureTtsProvider } from './azure.js';
+import { createCustomTtsProvider } from './custom.js';
+import { createEdgeTtsProvider } from './edge.js';
+import { createOpenAiTtsProvider } from './openai.js';
+export function resolveTtsProvider(config, override) {
+    const ttsConfig = config.tts || {};
+    const id = override || ttsConfig.provider || 'edge';
+    switch (id) {
+        case 'openai':
+            return createOpenAiTtsProvider(config);
+        case 'azure':
+            return createAzureTtsProvider(ttsConfig);
+        case 'custom':
+            return createCustomTtsProvider(config);
+        case 'edge':
+        default:
+            return createEdgeTtsProvider(ttsConfig);
+    }
+}
+//# sourceMappingURL=index.js.map

package/lib/tts/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/tts/index.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,sBAAsB,EAAE,MAAM,YAAY,CAAC;AACpD,OAAO,EAAE,uBAAuB,EAAE,MAAM,aAAa,CAAC;AACtD,OAAO,EAAE,qBAAqB,EAAE,MAAM,WAAW,CAAC;AAClD,OAAO,EAAE,uBAAuB,EAAE,MAAM,aAAa,CAAC;AAEtD,MAAM,UAAU,kBAAkB,CAAC,MAAoB,EAAE,QAAwB;IAC/E,MAAM,SAAS,GAAG,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC;IACnC,MAAM,EAAE,GAAG,QAAQ,IAAI,SAAS,CAAC,QAAQ,IAAI,MAAM,CAAC;IAEpD,QAAQ,EAAE,EAAE,CAAC;QACX,KAAK,QAAQ;YACX,OAAO,uBAAuB,CAAC,MAAM,CAAC,CAAC;QACzC,KAAK,OAAO;YACV,OAAO,sBAAsB,CAAC,SAAS,CAAC,CAAC;QAC3C,KAAK,QAAQ;YACX,OAAO,uBAAuB,CAAC,MAAM,CAAC,CAAC;QACzC,KAAK,MAAM,CAAC;QACZ;YACE,OAAO,qBAAqB,CAAC,SAAS,CAAC,CAAC;IAC5C,CAAC;AACH,CAAC"}

package/lib/tts/openai.d.ts

@@ -0,0 +1,3 @@
+import type { SpeechConfig, TtsProvider } from '../types.js';
+export declare function createOpenAiTtsProvider(config: SpeechConfig): TtsProvider;
+//# sourceMappingURL=openai.d.ts.map

package/lib/tts/openai.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"openai.d.ts","sourceRoot":"","sources":["../../src/tts/openai.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAa,WAAW,EAA2C,MAAM,aAAa,CAAC;AAMjH,wBAAgB,uBAAuB,CAAC,MAAM,EAAE,YAAY,GAAG,WAAW,CAqCzE"}

package/lib/tts/openai.js

@@ -0,0 +1,40 @@
+function resolveOpenAiApiKey(config) {
+    return config.tts?.apiKey || config.stt?.apiKey || '';
+}
+export function createOpenAiTtsProvider(config) {
+    const ttsConfig = config.tts || {};
+    return {
+        id: 'openai',
+        async synthesize(input) {
+            const host = (ttsConfig.host || 'https://api.openai.com').replace(/\/$/, '');
+            const model = ttsConfig.model || 'tts-1';
+            const voice = input.voice || ttsConfig.voice || 'alloy';
+            const apiKey = resolveOpenAiApiKey(config);
+            const format = input.format === 'wav' ? 'wav' : 'mp3';
+            const body = {
+                model,
+                input: input.text,
+                voice,
+                response_format: format,
+            };
+            if (input.speed != null)
+                body.speed = input.speed;
+            else if (ttsConfig.speed != null)
+                body.speed = ttsConfig.speed;
+            const response = await fetch(`${host}/v1/audio/speech`, {
+                method: 'POST',
+                headers: {
+                    Authorization: `Bearer ${apiKey}`,
+                    'Content-Type': 'application/json',
+                },
+                body: JSON.stringify(body),
+            });
+            if (!response.ok) {
+                throw new Error(`OpenAI TTS failed: ${response.status} ${response.statusText}`);
+            }
+            const data = Buffer.from(await response.arrayBuffer());
+            return { data, format };
+        },
+    };
+}
+//# sourceMappingURL=openai.js.map

package/lib/tts/openai.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"openai.js","sourceRoot":"","sources":["../../src/tts/openai.ts"],"names":[],"mappings":"AAEA,SAAS,mBAAmB,CAAC,MAAoB;IAC/C,OAAO,MAAM,CAAC,GAAG,EAAE,MAAM,IAAI,MAAM,CAAC,GAAG,EAAE,MAAM,IAAI,EAAE,CAAC;AACxD,CAAC;AAED,MAAM,UAAU,uBAAuB,CAAC,MAAoB;IAC1D,MAAM,SAAS,GAAG,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC;IACnC,OAAO;QACL,EAAE,EAAE,QAAQ;QACZ,KAAK,CAAC,UAAU,CAAC,KAAyB;YACxC,MAAM,IAAI,GAAG,CAAC,SAAS,CAAC,IAAI,IAAI,wBAAwB,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;YAC7E,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,IAAI,OAAO,CAAC;YACzC,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,IAAI,SAAS,CAAC,KAAK,IAAI,OAAO,CAAC;YACxD,MAAM,MAAM,GAAG,mBAAmB,CAAC,MAAM,CAAC,CAAC;YAC3C,MAAM,MAAM,GAAG,KAAK,CAAC,MAAM,KAAK,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC;YAEtD,MAAM,IAAI,GAA4B;gBACpC,KAAK;gBACL,KAAK,EAAE,KAAK,CAAC,IAAI;gBACjB,KAAK;gBACL,eAAe,EAAE,MAAM;aACxB,CAAC;YACF,IAAI,KAAK,CAAC,KAAK,IAAI,IAAI;gBAAE,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC;iBAC7C,IAAI,SAAS,CAAC,KAAK,IAAI,IAAI;gBAAE,IAAI,CAAC,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC;YAE/D,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,GAAG,IAAI,kBAAkB,EAAE;gBACtD,MAAM,EAAE,MAAM;gBACd,OAAO,EAAE;oBACP,aAAa,EAAE,UAAU,MAAM,EAAE;oBACjC,cAAc,EAAE,kBAAkB;iBACnC;gBACD,IAAI,EAAE,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC;aAC3B,CAAC,CAAC;YAEH,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;gBACjB,MAAM,IAAI,KAAK,CAAC,sBAAsB,QAAQ,CAAC,MAAM,IAAI,QAAQ,CAAC,UAAU,EAAE,CAAC,CAAC;YAClF,CAAC;YAED,MAAM,IAAI,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,QAAQ,CAAC,WAAW,EAAE,CAAC,CAAC;YACvD,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC;QAC1B,CAAC;KACF,CAAC;AACJ,CAAC"}

package/lib/types.d.ts

@@ -0,0 +1,60 @@
+export type SttProviderId = 'ollama' | 'openai';
+export type TtsProviderId = 'edge' | 'openai' | 'azure' | 'custom';
+export interface STTConfig {
+    enabled?: boolean;
+    provider?: SttProviderId;
+    model?: string;
+    host?: string;
+    apiKey?: string;
+}
+export interface TTSConfig {
+    enabled?: boolean;
+    provider?: TtsProviderId;
+    voice?: string;
+    model?: string;
+    host?: string;
+    apiKey?: string;
+    /** edge-tts */
+    rate?: string;
+    pitch?: string;
+    edgeTtsCommand?: string;
+    /** azure */
+    region?: string;
+    subscriptionKey?: string;
+    /** custom OpenAI-compatible */
+    baseUrl?: string;
+    headers?: Record<string, string>;
+    speed?: number;
+}
+export interface SpeechConfig {
+    stt?: STTConfig;
+    tts?: TTSConfig;
+}
+export interface TtsSynthesizeInput {
+    text: string;
+    voice?: string;
+    format?: 'mp3' | 'wav';
+    speed?: number;
+    provider?: TtsProviderId;
+}
+export interface TtsSynthesizeResult {
+    data: Buffer;
+    format: 'mp3' | 'wav';
+}
+export interface TtsProvider {
+    readonly id: TtsProviderId;
+    synthesize(input: TtsSynthesizeInput): Promise<TtsSynthesizeResult>;
+}
+export interface TranscribeInput {
+    data: Buffer;
+    mimeType?: string;
+}
+export interface SpeechPipeline {
+    transcribe(input: TranscribeInput): Promise<string>;
+    synthesize(input: TtsSynthesizeInput): Promise<TtsSynthesizeResult>;
+}
+export interface SpeechLogger {
+    debug?(message: string, ...args: unknown[]): void;
+    warn?(message: string, ...args: unknown[]): void;
+}
+//# sourceMappingURL=types.d.ts.map

package/lib/types.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../src/types.ts"],"names":[],"mappings":"AAAA,MAAM,MAAM,aAAa,GAAG,QAAQ,GAAG,QAAQ,CAAC;AAChD,MAAM,MAAM,aAAa,GAAG,MAAM,GAAG,QAAQ,GAAG,OAAO,GAAG,QAAQ,CAAC;AAEnE,MAAM,WAAW,SAAS;IACxB,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB,QAAQ,CAAC,EAAE,aAAa,CAAC;IACzB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AAED,MAAM,WAAW,SAAS;IACxB,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB,QAAQ,CAAC,EAAE,aAAa,CAAC;IACzB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,eAAe;IACf,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,YAAY;IACZ,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,+BAA+B;IAC/B,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IACjC,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED,MAAM,WAAW,YAAY;IAC3B,GAAG,CAAC,EAAE,SAAS,CAAC;IAChB,GAAG,CAAC,EAAE,SAAS,CAAC;CACjB;AAED,MAAM,WAAW,kBAAkB;IACjC,IAAI,EAAE,MAAM,CAAC;IACb,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,MAAM,CAAC,EAAE,KAAK,GAAG,KAAK,CAAC;IACvB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,aAAa,CAAC;CAC1B;AAED,MAAM,WAAW,mBAAmB;IAClC,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,KAAK,GAAG,KAAK,CAAC;CACvB;AAED,MAAM,WAAW,WAAW;IAC1B,QAAQ,CAAC,EAAE,EAAE,aAAa,CAAC;IAC3B,UAAU,CAAC,KAAK,EAAE,kBAAkB,GAAG,OAAO,CAAC,mBAAmB,CAAC,CAAC;CACrE;AAED,MAAM,WAAW,eAAe;IAC9B,IAAI,EAAE,MAAM,CAAC;IACb,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,MAAM,WAAW,cAAc;IAC7B,UAAU,CAAC,KAAK,EAAE,eAAe,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC;IACpD,UAAU,CAAC,KAAK,EAAE,kBAAkB,GAAG,OAAO,CAAC,mBAAmB,CAAC,CAAC;CACrE;AAED,MAAM,WAAW,YAAY;IAC3B,KAAK,CAAC,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,IAAI,EAAE,OAAO,EAAE,GAAG,IAAI,CAAC;IAClD,IAAI,CAAC,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,IAAI,EAAE,OAAO,EAAE,GAAG,IAAI,CAAC;CAClD"}

package/lib/types.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=types.js.map

package/lib/types.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.js","sourceRoot":"","sources":["../src/types.ts"],"names":[],"mappings":""}

package/package.json

@@ -0,0 +1,62 @@
+{
+  "name": "@zhin.js/speech",
+  "version": "0.0.1",
+  "description": "STT/TTS for Zhin.js (Whisper + edge/openai/azure/custom TTS); optional peer of zhin.js",
+  "type": "module",
+  "main": "./lib/index.js",
+  "types": "./lib/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./lib/index.d.ts",
+      "development": "./src/index.ts",
+      "import": "./lib/index.js"
+    },
+    "./package.json": "./package.json"
+  },
+  "files": [
+    "src",
+    "lib",
+    "README.md"
+  ],
+  "keywords": [
+    "zhin",
+    "zhin.js",
+    "speech",
+    "stt",
+    "tts",
+    "whisper",
+    "edge-tts"
+  ],
+  "author": {
+    "name": "lc-cn",
+    "email": "admin@liucl.cn",
+    "url": "https://github.com/lc-cn"
+  },
+  "license": "MIT",
+  "peerDependencies": {
+    "@zhin.js/core": "1.3.1"
+  },
+  "peerDependenciesMeta": {
+    "@zhin.js/core": {
+      "optional": true
+    }
+  },
+  "devDependencies": {
+    "typescript": "^6.0.3",
+    "@zhin.js/core": "1.3.1"
+  },
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/zhinjs/zhin.git",
+    "directory": "packages/toolkit/speech"
+  },
+  "publishConfig": {
+    "access": "public",
+    "registry": "https://registry.npmjs.org"
+  },
+  "scripts": {
+    "build": "tsc --build",
+    "clean": "rimraf lib",
+    "test": "vitest run"
+  }
+}

package/src/index.ts

@@ -0,0 +1,18 @@
+export { createSpeechPipeline } from './pipeline.js';
+export { resolveTtsProvider } from './tts/index.js';
+export type {
+  SpeechConfig,
+  SpeechLogger,
+  SpeechPipeline,
+  STTConfig,
+  TTSConfig,
+  TranscribeInput,
+  TtsProvider,
+  TtsProviderId,
+  TtsSynthesizeInput,
+  TtsSynthesizeResult,
+  SttProviderId,
+} from './types.js';
+
+/** 动态 import 时使用的包名（与 package.json name 一致） */
+export const SPEECH_PACKAGE = '@zhin.js/speech';

package/src/pipeline.ts

@@ -0,0 +1,67 @@
+import { transcribeWithOllama } from './stt/ollama.js';
+import { transcribeWithOpenAI } from './stt/openai.js';
+import { resolveTtsProvider } from './tts/index.js';
+import type {
+  SpeechConfig,
+  SpeechLogger,
+  SpeechPipeline,
+  TranscribeInput,
+  TtsSynthesizeInput,
+} from './types.js';
+
+const DEFAULT_SPEECH_CONFIG: SpeechConfig = {
+  stt: {
+    enabled: true,
+    provider: 'ollama',
+    model: 'whisper',
+    host: 'http://localhost:11434',
+  },
+  tts: {
+    enabled: true,
+    provider: 'edge',
+    voice: 'zh-CN-XiaoxiaoNeural',
+    rate: '+0%',
+    pitch: '+0Hz',
+    edgeTtsCommand: 'edge-tts',
+  },
+};
+
+function mergeSpeechConfig(config?: SpeechConfig): SpeechConfig {
+  return {
+    stt: { ...DEFAULT_SPEECH_CONFIG.stt, ...config?.stt },
+    tts: { ...DEFAULT_SPEECH_CONFIG.tts, ...config?.tts },
+  };
+}
+
+export function createSpeechPipeline(
+  config?: SpeechConfig,
+  _logger?: SpeechLogger,
+): SpeechPipeline {
+  const merged = mergeSpeechConfig(config);
+  const sttConfig = merged.stt || {};
+  const defaultTts = resolveTtsProvider(merged);
+
+  return {
+    async transcribe(input: TranscribeInput): Promise<string> {
+      if (sttConfig.enabled === false) {
+        throw new Error('STT is disabled in speech config');
+      }
+      const provider = sttConfig.provider || 'ollama';
+      const mimeType = input.mimeType || 'audio/wav';
+      if (provider === 'openai') {
+        return transcribeWithOpenAI(input.data, sttConfig, mimeType);
+      }
+      return transcribeWithOllama(input.data, sttConfig, mimeType);
+    },
+
+    async synthesize(input: TtsSynthesizeInput) {
+      if (merged.tts?.enabled === false) {
+        throw new Error('TTS is disabled in speech config');
+      }
+      const provider = input.provider
+        ? resolveTtsProvider(merged, input.provider)
+        : defaultTts;
+      return provider.synthesize(input);
+    },
+  };
+}

package/src/stt/ollama.ts

@@ -0,0 +1,32 @@
+import type { STTConfig } from '../types.js';
+
+export async function transcribeWithOllama(
+  audioData: Buffer,
+  sttConfig: STTConfig,
+  _mimeType: string = 'audio/wav',
+): Promise<string> {
+  const host = sttConfig.host || 'http://localhost:11434';
+  const model = sttConfig.model || 'whisper';
+  const base64Audio = audioData.toString('base64');
+
+  const response = await fetch(`${host}/api/chat`, {
+    method: 'POST',
+    headers: { 'Content-Type': 'application/json' },
+    body: JSON.stringify({
+      model,
+      messages: [{
+        role: 'user',
+        content: '请将这段音频转写为文字，只输出转写的文字内容，不要加任何说明。',
+        images: [base64Audio],
+      }],
+      stream: false,
+    }),
+  });
+
+  if (!response.ok) {
+    throw new Error(`Ollama STT failed: ${response.status} ${response.statusText}`);
+  }
+
+  const data = await response.json() as { message?: { content?: string } };
+  return data.message?.content?.trim() || '';
+}

package/src/stt/openai.ts

@@ -0,0 +1,42 @@
+import { randomBytes } from 'node:crypto';
+import { tmpdir } from 'node:os';
+import { join } from 'node:path';
+import { unlink, writeFile } from 'node:fs/promises';
+import type { STTConfig } from '../types.js';
+
+export async function transcribeWithOpenAI(
+  audioData: Buffer,
+  sttConfig: STTConfig,
+  mimeType: string = 'audio/wav',
+): Promise<string> {
+  const host = sttConfig.host || 'https://api.openai.com';
+  const model = sttConfig.model || 'whisper-1';
+  const apiKey = sttConfig.apiKey || '';
+
+  const tmpFile = join(tmpdir(), `zhin-stt-${randomBytes(8).toString('hex')}.wav`);
+  await writeFile(tmpFile, audioData);
+
+  try {
+    const ext = mimeType.includes('mp3') ? 'mp3' : 'wav';
+    const formData = new FormData();
+    const blob = new Blob([new Uint8Array(audioData)], { type: mimeType });
+    formData.append('file', blob, `audio.${ext}`);
+    formData.append('model', model);
+    formData.append('language', 'zh');
+
+    const response = await fetch(`${host}/v1/audio/transcriptions`, {
+      method: 'POST',
+      headers: { Authorization: `Bearer ${apiKey}` },
+      body: formData,
+    });
+
+    if (!response.ok) {
+      throw new Error(`OpenAI STT failed: ${response.status} ${response.statusText}`);
+    }
+
+    const data = await response.json() as { text?: string };
+    return data.text?.trim() || '';
+  } finally {
+    await unlink(tmpFile).catch(() => {});
+  }
+}

package/src/tts/azure.ts

@@ -0,0 +1,42 @@
+import type { TTSConfig, TtsProvider, TtsSynthesizeInput, TtsSynthesizeResult } from '../types.js';
+
+/** Azure Cognitive Services text-to-speech REST (output format: audio-16khz-128kbitrate-mono-mp3) */
+export function createAzureTtsProvider(config: TTSConfig): TtsProvider {
+  return {
+    id: 'azure',
+    async synthesize(input: TtsSynthesizeInput): Promise<TtsSynthesizeResult> {
+      const region = config.region || 'eastasia';
+      const subscriptionKey = config.subscriptionKey || '';
+      const voice = input.voice || config.voice || 'zh-CN-XiaoxiaoNeural';
+      const url = `https://${region}.tts.speech.microsoft.com/cognitiveservices/v1`;
+
+      const ssml = `<speak version='1.0' xml:lang='zh-CN'><voice name='${voice}'>${escapeXml(input.text)}</voice></speak>`;
+
+      const response = await fetch(url, {
+        method: 'POST',
+        headers: {
+          'Ocp-Apim-Subscription-Key': subscriptionKey,
+          'Content-Type': 'application/ssml+xml',
+          'X-Microsoft-OutputFormat': 'audio-16khz-128kbitrate-mono-mp3',
+        },
+        body: ssml,
+      });
+
+      if (!response.ok) {
+        throw new Error(`Azure TTS failed: ${response.status} ${response.statusText}`);
+      }
+
+      const data = Buffer.from(await response.arrayBuffer());
+      return { data, format: 'mp3' };
+    },
+  };
+}
+
+function escapeXml(text: string): string {
+  return text
+    .replace(/&/g, '&amp;')
+    .replace(/</g, '&lt;')
+    .replace(/>/g, '&gt;')
+    .replace(/"/g, '&quot;')
+    .replace(/'/g, '&apos;');
+}

package/src/tts/custom.ts

@@ -0,0 +1,50 @@
+import type { SpeechConfig, TTSConfig, TtsProvider, TtsSynthesizeInput, TtsSynthesizeResult } from '../types.js';
+
+function resolveOpenAiApiKey(config: SpeechConfig): string {
+  return config.tts?.apiKey || config.stt?.apiKey || '';
+}
+
+export function createCustomTtsProvider(config: SpeechConfig): TtsProvider {
+  const ttsConfig = config.tts || {};
+  return {
+    id: 'custom',
+    async synthesize(input: TtsSynthesizeInput): Promise<TtsSynthesizeResult> {
+      const baseUrl = (ttsConfig.baseUrl || '').replace(/\/$/, '');
+      if (!baseUrl) {
+        throw new Error('custom TTS requires speech.tts.baseUrl');
+      }
+      const model = ttsConfig.model || 'tts-1';
+      const voice = input.voice || ttsConfig.voice || 'default';
+      const apiKey = resolveOpenAiApiKey(config);
+      const format = input.format === 'wav' ? 'wav' : 'mp3';
+
+      const headers: Record<string, string> = {
+        'Content-Type': 'application/json',
+        ...(ttsConfig.headers || {}),
+      };
+      if (apiKey) headers.Authorization = `Bearer ${apiKey}`;
+
+      const body: Record<string, unknown> = {
+        model,
+        input: input.text,
+        voice,
+        response_format: format,
+      };
+      if (input.speed != null) body.speed = input.speed;
+      else if (ttsConfig.speed != null) body.speed = ttsConfig.speed;
+
+      const response = await fetch(baseUrl, {
+        method: 'POST',
+        headers,
+        body: JSON.stringify(body),
+      });
+
+      if (!response.ok) {
+        throw new Error(`Custom TTS failed: ${response.status} ${response.statusText}`);
+      }
+
+      const data = Buffer.from(await response.arrayBuffer());
+      return { data, format };
+    },
+  };
+}

package/src/tts/edge.ts

@@ -0,0 +1,34 @@
+import { exec } from 'node:child_process';
+import { randomBytes } from 'node:crypto';
+import { readFile, unlink } from 'node:fs/promises';
+import { tmpdir } from 'node:os';
+import { join } from 'node:path';
+import { promisify } from 'node:util';
+import type { TTSConfig, TtsProvider, TtsSynthesizeInput, TtsSynthesizeResult } from '../types.js';
+
+const execAsync = promisify(exec);
+
+export function createEdgeTtsProvider(config: TTSConfig): TtsProvider {
+  return {
+    id: 'edge',
+    async synthesize(input: TtsSynthesizeInput): Promise<TtsSynthesizeResult> {
+      const voice = input.voice || config.voice || 'zh-CN-XiaoxiaoNeural';
+      const rate = config.rate || '+0%';
+      const pitch = config.pitch || '+0Hz';
+      const cmd = config.edgeTtsCommand || 'edge-tts';
+      const tmpFile = join(tmpdir(), `zhin-tts-${randomBytes(8).toString('hex')}.mp3`);
+      const safeText = input.text.replace(/"/g, '\\"').replace(/'/g, "\\'");
+
+      try {
+        await execAsync(
+          `${cmd} --voice "${voice}" --rate="${rate}" --pitch="${pitch}" --text "${safeText}" --write-media "${tmpFile}"`,
+          { timeout: 30_000 },
+        );
+        const data = await readFile(tmpFile);
+        return { data, format: 'mp3' };
+      } finally {
+        await unlink(tmpFile).catch(() => {});
+      }
+    },
+  };
+}

package/src/tts/index.ts

@@ -0,0 +1,22 @@
+import type { SpeechConfig, TtsProvider, TtsProviderId } from '../types.js';
+import { createAzureTtsProvider } from './azure.js';
+import { createCustomTtsProvider } from './custom.js';
+import { createEdgeTtsProvider } from './edge.js';
+import { createOpenAiTtsProvider } from './openai.js';
+
+export function resolveTtsProvider(config: SpeechConfig, override?: TtsProviderId): TtsProvider {
+  const ttsConfig = config.tts || {};
+  const id = override || ttsConfig.provider || 'edge';
+
+  switch (id) {
+    case 'openai':
+      return createOpenAiTtsProvider(config);
+    case 'azure':
+      return createAzureTtsProvider(ttsConfig);
+    case 'custom':
+      return createCustomTtsProvider(config);
+    case 'edge':
+    default:
+      return createEdgeTtsProvider(ttsConfig);
+  }
+}

package/src/tts/openai.ts

@@ -0,0 +1,44 @@
+import type { SpeechConfig, TTSConfig, TtsProvider, TtsSynthesizeInput, TtsSynthesizeResult } from '../types.js';
+
+function resolveOpenAiApiKey(config: SpeechConfig): string {
+  return config.tts?.apiKey || config.stt?.apiKey || '';
+}
+
+export function createOpenAiTtsProvider(config: SpeechConfig): TtsProvider {
+  const ttsConfig = config.tts || {};
+  return {
+    id: 'openai',
+    async synthesize(input: TtsSynthesizeInput): Promise<TtsSynthesizeResult> {
+      const host = (ttsConfig.host || 'https://api.openai.com').replace(/\/$/, '');
+      const model = ttsConfig.model || 'tts-1';
+      const voice = input.voice || ttsConfig.voice || 'alloy';
+      const apiKey = resolveOpenAiApiKey(config);
+      const format = input.format === 'wav' ? 'wav' : 'mp3';
+
+      const body: Record<string, unknown> = {
+        model,
+        input: input.text,
+        voice,
+        response_format: format,
+      };
+      if (input.speed != null) body.speed = input.speed;
+      else if (ttsConfig.speed != null) body.speed = ttsConfig.speed;
+
+      const response = await fetch(`${host}/v1/audio/speech`, {
+        method: 'POST',
+        headers: {
+          Authorization: `Bearer ${apiKey}`,
+          'Content-Type': 'application/json',
+        },
+        body: JSON.stringify(body),
+      });
+
+      if (!response.ok) {
+        throw new Error(`OpenAI TTS failed: ${response.status} ${response.statusText}`);
+      }
+
+      const data = Buffer.from(await response.arrayBuffer());
+      return { data, format };
+    },
+  };
+}

package/src/types.ts

@@ -0,0 +1,68 @@
+export type SttProviderId = 'ollama' | 'openai';
+export type TtsProviderId = 'edge' | 'openai' | 'azure' | 'custom';
+
+export interface STTConfig {
+  enabled?: boolean;
+  provider?: SttProviderId;
+  model?: string;
+  host?: string;
+  apiKey?: string;
+}
+
+export interface TTSConfig {
+  enabled?: boolean;
+  provider?: TtsProviderId;
+  voice?: string;
+  model?: string;
+  host?: string;
+  apiKey?: string;
+  /** edge-tts */
+  rate?: string;
+  pitch?: string;
+  edgeTtsCommand?: string;
+  /** azure */
+  region?: string;
+  subscriptionKey?: string;
+  /** custom OpenAI-compatible */
+  baseUrl?: string;
+  headers?: Record<string, string>;
+  speed?: number;
+}
+
+export interface SpeechConfig {
+  stt?: STTConfig;
+  tts?: TTSConfig;
+}
+
+export interface TtsSynthesizeInput {
+  text: string;
+  voice?: string;
+  format?: 'mp3' | 'wav';
+  speed?: number;
+  provider?: TtsProviderId;
+}
+
+export interface TtsSynthesizeResult {
+  data: Buffer;
+  format: 'mp3' | 'wav';
+}
+
+export interface TtsProvider {
+  readonly id: TtsProviderId;
+  synthesize(input: TtsSynthesizeInput): Promise<TtsSynthesizeResult>;
+}
+
+export interface TranscribeInput {
+  data: Buffer;
+  mimeType?: string;
+}
+
+export interface SpeechPipeline {
+  transcribe(input: TranscribeInput): Promise<string>;
+  synthesize(input: TtsSynthesizeInput): Promise<TtsSynthesizeResult>;
+}
+
+export interface SpeechLogger {
+  debug?(message: string, ...args: unknown[]): void;
+  warn?(message: string, ...args: unknown[]): void;
+}