@amaster.ai/tts-client 1.1.0-beta.51 → 1.1.0-beta.53

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (7) hide show
  1. package/dist/index.cjs +15 -5
  2. package/dist/index.cjs.map +1 -1
  3. package/dist/index.d.cts +2 -0
  4. package/dist/index.d.ts +2 -0
  5. package/dist/index.js +15 -5
  6. package/dist/index.js.map +1 -1
  7. package/package.json +1 -1
package/dist/index.cjs CHANGED
@@ -42,6 +42,7 @@ function createTTSClient(config) {
42
42
  let ws = null;
43
43
  let audioChunks = [];
44
44
  let audioContext = null;
45
+ let audioSource = null;
45
46
  async function connect() {
46
47
  return new Promise((resolve, reject) => {
47
48
  let wsUrl = TTS_PATH;
@@ -160,25 +161,34 @@ function createTTSClient(config) {
160
161
  source.connect(audioContext.destination);
161
162
  source.onended = () => onAudioEnd?.();
162
163
  source.start(0);
164
+ audioSource = source;
163
165
  } catch (err) {
164
166
  onError?.(err);
165
167
  }
166
168
  }
167
- function close() {
168
- if (ws) {
169
- ws.close();
170
- ws = null;
169
+ function stopAudio() {
170
+ if (audioSource) {
171
+ audioSource.stop();
172
+ audioSource = null;
171
173
  }
172
174
  if (audioContext) {
173
175
  audioContext.close();
174
176
  audioContext = null;
175
177
  }
176
178
  }
179
+ function close() {
180
+ if (ws) {
181
+ ws.close();
182
+ ws = null;
183
+ }
184
+ stopAudio();
185
+ }
177
186
  return {
178
187
  connect,
179
188
  speak,
180
189
  close,
181
- play: playAudio
190
+ play: playAudio,
191
+ stop: stopAudio
182
192
  };
183
193
  }
184
194
  var tts_client_default = (authConfig) => {
package/dist/index.cjs.map CHANGED
@@ -1 +1 @@
1
- {"version":3,"sources":["../src/index.ts","../src/tts-client.ts"],"sourcesContent":["import type { TTSClient, TTSClientConfig } from \"./tts-client\";\nimport createTTSClient from \"./tts-client\";\nexport { createTTSClient, type TTSClient, type TTSClientConfig };\n","/**\n * TTS Realtime WebSocket Client\n *\n * WebSocket-based real-time text-to-speech synthesis with multiple voice options.\n * Built-in playback only supports PCM format.\n *\n * @example\n * ```typescript\n * const client = createTTSClient({\n * voice: \"Cherry\",\n * autoPlay: true,\n * onReady() {\n * console.log(\"TTS ready\");\n * },\n * onAudioStart() {\n * console.log(\"Playing audio\");\n * },\n * onAudioEnd() {\n * console.log(\"Playback ended\");\n * },\n * });\n *\n * await client.connect();\n * await client.speak(\"Hello, this is a test.\");\n * // client.close();\n * ```\n */\n\nconst TTS_PATH = \"/api/proxy/builtin/platform/qwen-tts/api-ws/v1/realtime\";\n\nexport interface TTSClientConfig {\n /** Get access token for WebSocket authentication */\n getAccessToken?: () => string | null;\n /** Voice name, default 'Cherry' */\n voice?: string;\n /** Auto play audio, default true */\n autoPlay?: boolean;\n /** Audio format, default 'pcm' */\n audioFormat?: \"pcm\" | \"mp3\" | \"wav\" | \"opus\";\n /** Sample rate, default 24000 */\n sampleRate?: number;\n /** Called when connection is ready */\n onReady?: () => void;\n /** Called when audio playback starts */\n onAudioStart?: () => void;\n /** Called when audio playback ends */\n onAudioEnd?: () => void;\n /** Called on each audio chunk received */\n onAudioChunk?: (chunk: string[]) => void;\n /** Called on error */\n onError?: (error: Error) => void;\n}\n\nexport interface TTSClient {\n /** Connect to TTS service */\n connect(): Promise<void>;\n /** Synthesize speech from text */\n speak(text: string): Promise<void>;\n /** Play audio from chunks */\n play(): void;\n /** Close connection */\n close(): void;\n}\n\nfunction createTTSClient(config: TTSClientConfig): TTSClient {\n const {\n getAccessToken,\n voice = \"Cherry\",\n autoPlay = true,\n audioFormat = \"pcm\",\n sampleRate = 24000,\n onReady,\n onAudioStart,\n onAudioEnd,\n onAudioChunk,\n onError,\n } = config;\n\n let ws: WebSocket | null = null;\n let audioChunks: string[] = [];\n let audioContext: AudioContext | null = null;\n\n async function connect(): Promise<void> {\n return new Promise((resolve, reject) => {\n // Build WebSocket URL with optional token parameter\n let wsUrl = TTS_PATH;\n if (getAccessToken) {\n const token = getAccessToken();\n if (token) {\n const separator = wsUrl.includes(\"?\") ? \"&\" : \"?\";\n wsUrl = `${wsUrl}${separator}token=${encodeURIComponent(token)}`;\n }\n }\n\n ws = new WebSocket(wsUrl);\n\n ws.onopen = () => {};\n\n ws.onmessage = (event) => {\n const data = JSON.parse(event.data);\n\n if (data.type === \"session.created\") {\n ws!.send(\n JSON.stringify({\n type: \"session.update\",\n session: {\n mode: \"server_commit\",\n voice,\n response_format: audioFormat,\n sample_rate: sampleRate,\n },\n })\n );\n }\n\n if (data.type === \"session.updated\") {\n onReady?.();\n resolve();\n }\n\n if (data.type === \"response.audio.delta\") {\n audioChunks.push(data.delta);\n onAudioChunk?.(audioChunks);\n }\n\n if (data.type === \"response.audio.done\") {\n onAudioChunk?.(audioChunks);\n if (autoPlay && typeof window !== \"undefined\") {\n playAudio();\n }\n }\n\n if (data.type === \"response.done\") {\n ws!.send(JSON.stringify({ type: \"session.finish\" }));\n }\n\n if (data.type === \"error\") {\n const err = new Error(data.error?.message || \"Unknown error\");\n onError?.(err);\n reject(err);\n }\n };\n\n ws.onerror = () => {\n const err = new Error(\"WebSocket connection error\");\n onError?.(err);\n reject(err);\n };\n\n ws.onclose = () => {\n ws = null;\n };\n });\n }\n\n async function speak(text: string): Promise<void> {\n if (!ws || ws.readyState !== WebSocket.OPEN) {\n throw new Error(\"WebSocket not connected\");\n }\n\n audioChunks = [];\n\n ws.send(\n JSON.stringify({\n type: \"input_text_buffer.append\",\n text,\n })\n );\n\n setTimeout(() => {\n ws!.send(\n JSON.stringify({\n type: \"input_text_buffer.commit\",\n })\n );\n }, 100);\n }\n\n function playAudio() {\n let chunks: string[] = audioChunks;\n if (typeof window === \"undefined\") return;\n\n try {\n if (!audioContext) {\n audioContext = new AudioContext();\n }\n\n onAudioStart?.();\n\n let totalBytes = 0;\n const allBytes: Uint8Array[] = [];\n\n for (const chunk of chunks) {\n const binaryString = atob(chunk);\n const bytes = new Uint8Array(binaryString.length);\n for (let i = 0; i < binaryString.length; i++) {\n bytes[i] = binaryString.charCodeAt(i);\n }\n allBytes.push(bytes);\n totalBytes += bytes.length;\n }\n\n const combined = new Uint8Array(totalBytes);\n let offset = 0;\n for (const bytes of allBytes) {\n combined.set(bytes, offset);\n offset += bytes.length;\n }\n\n const numSamples = combined.length / 2;\n const audioBuffer = audioContext.createBuffer(1, numSamples, sampleRate);\n const channelData = audioBuffer.getChannelData(0);\n\n const dataView = new DataView(combined.buffer);\n for (let i = 0; i < numSamples; i++) {\n const int16 = dataView.getInt16(i * 2, true);\n channelData[i] = int16 / 32768.0;\n }\n\n const source = audioContext.createBufferSource();\n source.buffer = audioBuffer;\n source.connect(audioContext.destination);\n source.onended = () => onAudioEnd?.();\n source.start(0);\n } catch (err) {\n onError?.(err as Error);\n }\n }\n\n function close() {\n if (ws) {\n ws.close();\n ws = null;\n }\n if (audioContext) {\n audioContext.close();\n audioContext = null;\n }\n }\n\n return {\n connect,\n speak,\n close,\n play: playAudio,\n };\n}\n\nexport default (\n authConfig: Pick<TTSClientConfig, \"getAccessToken\">\n): ((config: TTSClientConfig) => TTSClient) => {\n return (config: TTSClientConfig) => {\n return createTTSClient({ ...authConfig, ...config });\n };\n};\n"],"mappings":";;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;;;AC4BA,IAAM,WAAW;AAoCjB,SAAS,gBAAgB,QAAoC;AAC3D,QAAM;AAAA,IACJ;AAAA,IACA,QAAQ;AAAA,IACR,WAAW;AAAA,IACX,cAAc;AAAA,IACd,aAAa;AAAA,IACb;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF,IAAI;AAEJ,MAAI,KAAuB;AAC3B,MAAI,cAAwB,CAAC;AAC7B,MAAI,eAAoC;AAExC,iBAAe,UAAyB;AACtC,WAAO,IAAI,QAAQ,CAAC,SAAS,WAAW;AAEtC,UAAI,QAAQ;AACZ,UAAI,gBAAgB;AAClB,cAAM,QAAQ,eAAe;AAC7B,YAAI,OAAO;AACT,gBAAM,YAAY,MAAM,SAAS,GAAG,IAAI,MAAM;AAC9C,kBAAQ,GAAG,KAAK,GAAG,SAAS,SAAS,mBAAmB,KAAK,CAAC;AAAA,QAChE;AAAA,MACF;AAEA,WAAK,IAAI,UAAU,KAAK;AAExB,SAAG,SAAS,MAAM;AAAA,MAAC;AAEnB,SAAG,YAAY,CAAC,UAAU;AACxB,cAAM,OAAO,KAAK,MAAM,MAAM,IAAI;AAElC,YAAI,KAAK,SAAS,mBAAmB;AACnC,aAAI;AAAA,YACF,KAAK,UAAU;AAAA,cACb,MAAM;AAAA,cACN,SAAS;AAAA,gBACP,MAAM;AAAA,gBACN;AAAA,gBACA,iBAAiB;AAAA,gBACjB,aAAa;AAAA,cACf;AAAA,YACF,CAAC;AAAA,UACH;AAAA,QACF;AAEA,YAAI,KAAK,SAAS,mBAAmB;AACnC,oBAAU;AACV,kBAAQ;AAAA,QACV;AAEA,YAAI,KAAK,SAAS,wBAAwB;AACxC,sBAAY,KAAK,KAAK,KAAK;AAC3B,yBAAe,WAAW;AAAA,QAC5B;AAEA,YAAI,KAAK,SAAS,uBAAuB;AACvC,yBAAe,WAAW;AAC1B,cAAI,YAAY,OAAO,WAAW,aAAa;AAC7C,sBAAU;AAAA,UACZ;AAAA,QACF;AAEA,YAAI,KAAK,SAAS,iBAAiB;AACjC,aAAI,KAAK,KAAK,UAAU,EAAE,MAAM,iBAAiB,CAAC,CAAC;AAAA,QACrD;AAEA,YAAI,KAAK,SAAS,SAAS;AACzB,gBAAM,MAAM,IAAI,MAAM,KAAK,OAAO,WAAW,eAAe;AAC5D,oBAAU,GAAG;AACb,iBAAO,GAAG;AAAA,QACZ;AAAA,MACF;AAEA,SAAG,UAAU,MAAM;AACjB,cAAM,MAAM,IAAI,MAAM,4BAA4B;AAClD,kBAAU,GAAG;AACb,eAAO,GAAG;AAAA,MACZ;AAEA,SAAG,UAAU,MAAM;AACjB,aAAK;AAAA,MACP;AAAA,IACF,CAAC;AAAA,EACH;AAEA,iBAAe,MAAM,MAA6B;AAChD,QAAI,CAAC,MAAM,GAAG,eAAe,UAAU,MAAM;AAC3C,YAAM,IAAI,MAAM,yBAAyB;AAAA,IAC3C;AAEA,kBAAc,CAAC;AAEf,OAAG;AAAA,MACD,KAAK,UAAU;AAAA,QACb,MAAM;AAAA,QACN;AAAA,MACF,CAAC;AAAA,IACH;AAEA,eAAW,MAAM;AACf,SAAI;AAAA,QACF,KAAK,UAAU;AAAA,UACb,MAAM;AAAA,QACR,CAAC;AAAA,MACH;AAAA,IACF,GAAG,GAAG;AAAA,EACR;AAEA,WAAS,YAAY;AACnB,QAAI,SAAmB;AACvB,QAAI,OAAO,WAAW,YAAa;AAEnC,QAAI;AACF,UAAI,CAAC,cAAc;AACjB,uBAAe,IAAI,aAAa;AAAA,MAClC;AAEA,qBAAe;AAEf,UAAI,aAAa;AACjB,YAAM,WAAyB,CAAC;AAEhC,iBAAW,SAAS,QAAQ;AAC1B,cAAM,eAAe,KAAK,KAAK;AAC/B,cAAM,QAAQ,IAAI,WAAW,aAAa,MAAM;AAChD,iBAAS,IAAI,GAAG,IAAI,aAAa,QAAQ,KAAK;AAC5C,gBAAM,CAAC,IAAI,aAAa,WAAW,CAAC;AAAA,QACtC;AACA,iBAAS,KAAK,KAAK;AACnB,sBAAc,MAAM;AAAA,MACtB;AAEA,YAAM,WAAW,IAAI,WAAW,UAAU;AAC1C,UAAI,SAAS;AACb,iBAAW,SAAS,UAAU;AAC5B,iBAAS,IAAI,OAAO,MAAM;AAC1B,kBAAU,MAAM;AAAA,MAClB;AAEA,YAAM,aAAa,SAAS,SAAS;AACrC,YAAM,cAAc,aAAa,aAAa,GAAG,YAAY,UAAU;AACvE,YAAM,cAAc,YAAY,eAAe,CAAC;AAEhD,YAAM,WAAW,IAAI,SAAS,SAAS,MAAM;AAC7C,eAAS,IAAI,GAAG,IAAI,YAAY,KAAK;AACnC,cAAM,QAAQ,SAAS,SAAS,IAAI,GAAG,IAAI;AAC3C,oBAAY,CAAC,IAAI,QAAQ;AAAA,MAC3B;AAEA,YAAM,SAAS,aAAa,mBAAmB;AAC/C,aAAO,SAAS;AAChB,aAAO,QAAQ,aAAa,WAAW;AACvC,aAAO,UAAU,MAAM,aAAa;AACpC,aAAO,MAAM,CAAC;AAAA,IAChB,SAAS,KAAK;AACZ,gBAAU,GAAY;AAAA,IACxB;AAAA,EACF;AAEA,WAAS,QAAQ;AACf,QAAI,IAAI;AACN,SAAG,MAAM;AACT,WAAK;AAAA,IACP;AACA,QAAI,cAAc;AAChB,mBAAa,MAAM;AACnB,qBAAe;AAAA,IACjB;AAAA,EACF;AAEA,SAAO;AAAA,IACL;AAAA,IACA;AAAA,IACA;AAAA,IACA,MAAM;AAAA,EACR;AACF;AAEA,IAAO,qBAAQ,CACb,eAC6C;AAC7C,SAAO,CAAC,WAA4B;AAClC,WAAO,gBAAgB,EAAE,GAAG,YAAY,GAAG,OAAO,CAAC;AAAA,EACrD;AACF;","names":[]}
1
+ {"version":3,"sources":["../src/index.ts","../src/tts-client.ts"],"sourcesContent":["import type { TTSClient, TTSClientConfig } from \"./tts-client\";\nimport createTTSClient from \"./tts-client\";\nexport { createTTSClient, type TTSClient, type TTSClientConfig };\n","/**\n * TTS Realtime WebSocket Client\n *\n * WebSocket-based real-time text-to-speech synthesis with multiple voice options.\n * Built-in playback only supports PCM format.\n *\n * @example\n * ```typescript\n * const client = createTTSClient({\n * voice: \"Cherry\",\n * autoPlay: true,\n * onReady() {\n * console.log(\"TTS ready\");\n * },\n * onAudioStart() {\n * console.log(\"Playing audio\");\n * },\n * onAudioEnd() {\n * console.log(\"Playback ended\");\n * },\n * });\n *\n * await client.connect();\n * await client.speak(\"Hello, this is a test.\");\n * // client.close();\n * ```\n */\n\nconst TTS_PATH = \"/api/proxy/builtin/platform/qwen-tts/api-ws/v1/realtime\";\n\nexport interface TTSClientConfig {\n /** Get access token for WebSocket authentication */\n getAccessToken?: () => string | null;\n /** Voice name, default 'Cherry' */\n voice?: string;\n /** Auto play audio, default true */\n autoPlay?: boolean;\n /** Audio format, default 'pcm' */\n audioFormat?: \"pcm\" | \"mp3\" | \"wav\" | \"opus\";\n /** Sample rate, default 24000 */\n sampleRate?: number;\n /** Called when connection is ready */\n onReady?: () => void;\n /** Called when audio playback starts */\n onAudioStart?: () => void;\n /** Called when audio playback ends */\n onAudioEnd?: () => void;\n /** Called on each audio chunk received */\n onAudioChunk?: (chunk: string[]) => void;\n /** Called on error */\n onError?: (error: Error) => void;\n}\n\nexport interface TTSClient {\n /** Connect to TTS service */\n connect(): Promise<void>;\n /** Synthesize speech from text */\n speak(text: string): Promise<void>;\n /** Play audio from chunks */\n play(): void;\n /** Stop audio playback */\n stop(): void;\n /** Close connection */\n close(): void;\n}\n\nfunction createTTSClient(config: TTSClientConfig): TTSClient {\n const {\n getAccessToken,\n voice = \"Cherry\",\n autoPlay = true,\n audioFormat = \"pcm\",\n sampleRate = 24000,\n onReady,\n onAudioStart,\n onAudioEnd,\n onAudioChunk,\n onError,\n } = config;\n\n let ws: WebSocket | null = null;\n let audioChunks: string[] = [];\n let audioContext: AudioContext | null = null;\n let audioSource: AudioBufferSourceNode | null = null;\n\n async function connect(): Promise<void> {\n return new Promise((resolve, reject) => {\n // Build WebSocket URL with optional token parameter\n let wsUrl = TTS_PATH;\n if (getAccessToken) {\n const token = getAccessToken();\n if (token) {\n const separator = wsUrl.includes(\"?\") ? \"&\" : \"?\";\n wsUrl = `${wsUrl}${separator}token=${encodeURIComponent(token)}`;\n }\n }\n\n ws = new WebSocket(wsUrl);\n\n ws.onopen = () => {};\n\n ws.onmessage = (event) => {\n const data = JSON.parse(event.data);\n\n if (data.type === \"session.created\") {\n ws!.send(\n JSON.stringify({\n type: \"session.update\",\n session: {\n mode: \"server_commit\",\n voice,\n response_format: audioFormat,\n sample_rate: sampleRate,\n },\n })\n );\n }\n\n if (data.type === \"session.updated\") {\n onReady?.();\n resolve();\n }\n\n if (data.type === \"response.audio.delta\") {\n audioChunks.push(data.delta);\n onAudioChunk?.(audioChunks);\n }\n\n if (data.type === \"response.audio.done\") {\n onAudioChunk?.(audioChunks);\n if (autoPlay && typeof window !== \"undefined\") {\n playAudio();\n }\n }\n\n if (data.type === \"response.done\") {\n ws!.send(JSON.stringify({ type: \"session.finish\" }));\n }\n\n if (data.type === \"error\") {\n const err = new Error(data.error?.message || \"Unknown error\");\n onError?.(err);\n reject(err);\n }\n };\n\n ws.onerror = () => {\n const err = new Error(\"WebSocket connection error\");\n onError?.(err);\n reject(err);\n };\n\n ws.onclose = () => {\n ws = null;\n };\n });\n }\n\n async function speak(text: string): Promise<void> {\n if (!ws || ws.readyState !== WebSocket.OPEN) {\n throw new Error(\"WebSocket not connected\");\n }\n\n audioChunks = [];\n\n ws.send(\n JSON.stringify({\n type: \"input_text_buffer.append\",\n text,\n })\n );\n\n setTimeout(() => {\n ws!.send(\n JSON.stringify({\n type: \"input_text_buffer.commit\",\n })\n );\n }, 100);\n }\n\n function playAudio() {\n let chunks: string[] = audioChunks;\n if (typeof window === \"undefined\") return;\n\n try {\n if (!audioContext) {\n audioContext = new AudioContext();\n }\n\n onAudioStart?.();\n\n let totalBytes = 0;\n const allBytes: Uint8Array[] = [];\n\n for (const chunk of chunks) {\n const binaryString = atob(chunk);\n const bytes = new Uint8Array(binaryString.length);\n for (let i = 0; i < binaryString.length; i++) {\n bytes[i] = binaryString.charCodeAt(i);\n }\n allBytes.push(bytes);\n totalBytes += bytes.length;\n }\n\n const combined = new Uint8Array(totalBytes);\n let offset = 0;\n for (const bytes of allBytes) {\n combined.set(bytes, offset);\n offset += bytes.length;\n }\n\n const numSamples = combined.length / 2;\n const audioBuffer = audioContext.createBuffer(1, numSamples, sampleRate);\n const channelData = audioBuffer.getChannelData(0);\n\n const dataView = new DataView(combined.buffer);\n for (let i = 0; i < numSamples; i++) {\n const int16 = dataView.getInt16(i * 2, true);\n channelData[i] = int16 / 32768.0;\n }\n\n const source = audioContext.createBufferSource();\n source.buffer = audioBuffer;\n source.connect(audioContext.destination);\n source.onended = () => onAudioEnd?.();\n source.start(0);\n audioSource = source;\n } catch (err) {\n onError?.(err as Error);\n }\n }\n\n function stopAudio() {\n if (audioSource) {\n audioSource.stop();\n audioSource = null;\n }\n if (audioContext) {\n audioContext.close();\n audioContext = null;\n }\n }\n\n function close() {\n if (ws) {\n ws.close();\n ws = null;\n }\n stopAudio();\n }\n\n return {\n connect,\n speak,\n close,\n play: playAudio,\n stop: stopAudio,\n };\n}\n\nexport default (\n authConfig: Pick<TTSClientConfig, \"getAccessToken\">\n): ((config: TTSClientConfig) => TTSClient) => {\n return (config: TTSClientConfig) => {\n return createTTSClient({ ...authConfig, ...config });\n };\n};\n"],"mappings":";;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;;;AC4BA,IAAM,WAAW;AAsCjB,SAAS,gBAAgB,QAAoC;AAC3D,QAAM;AAAA,IACJ;AAAA,IACA,QAAQ;AAAA,IACR,WAAW;AAAA,IACX,cAAc;AAAA,IACd,aAAa;AAAA,IACb;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF,IAAI;AAEJ,MAAI,KAAuB;AAC3B,MAAI,cAAwB,CAAC;AAC7B,MAAI,eAAoC;AACxC,MAAI,cAA4C;AAEhD,iBAAe,UAAyB;AACtC,WAAO,IAAI,QAAQ,CAAC,SAAS,WAAW;AAEtC,UAAI,QAAQ;AACZ,UAAI,gBAAgB;AAClB,cAAM,QAAQ,eAAe;AAC7B,YAAI,OAAO;AACT,gBAAM,YAAY,MAAM,SAAS,GAAG,IAAI,MAAM;AAC9C,kBAAQ,GAAG,KAAK,GAAG,SAAS,SAAS,mBAAmB,KAAK,CAAC;AAAA,QAChE;AAAA,MACF;AAEA,WAAK,IAAI,UAAU,KAAK;AAExB,SAAG,SAAS,MAAM;AAAA,MAAC;AAEnB,SAAG,YAAY,CAAC,UAAU;AACxB,cAAM,OAAO,KAAK,MAAM,MAAM,IAAI;AAElC,YAAI,KAAK,SAAS,mBAAmB;AACnC,aAAI;AAAA,YACF,KAAK,UAAU;AAAA,cACb,MAAM;AAAA,cACN,SAAS;AAAA,gBACP,MAAM;AAAA,gBACN;AAAA,gBACA,iBAAiB;AAAA,gBACjB,aAAa;AAAA,cACf;AAAA,YACF,CAAC;AAAA,UACH;AAAA,QACF;AAEA,YAAI,KAAK,SAAS,mBAAmB;AACnC,oBAAU;AACV,kBAAQ;AAAA,QACV;AAEA,YAAI,KAAK,SAAS,wBAAwB;AACxC,sBAAY,KAAK,KAAK,KAAK;AAC3B,yBAAe,WAAW;AAAA,QAC5B;AAEA,YAAI,KAAK,SAAS,uBAAuB;AACvC,yBAAe,WAAW;AAC1B,cAAI,YAAY,OAAO,WAAW,aAAa;AAC7C,sBAAU;AAAA,UACZ;AAAA,QACF;AAEA,YAAI,KAAK,SAAS,iBAAiB;AACjC,aAAI,KAAK,KAAK,UAAU,EAAE,MAAM,iBAAiB,CAAC,CAAC;AAAA,QACrD;AAEA,YAAI,KAAK,SAAS,SAAS;AACzB,gBAAM,MAAM,IAAI,MAAM,KAAK,OAAO,WAAW,eAAe;AAC5D,oBAAU,GAAG;AACb,iBAAO,GAAG;AAAA,QACZ;AAAA,MACF;AAEA,SAAG,UAAU,MAAM;AACjB,cAAM,MAAM,IAAI,MAAM,4BAA4B;AAClD,kBAAU,GAAG;AACb,eAAO,GAAG;AAAA,MACZ;AAEA,SAAG,UAAU,MAAM;AACjB,aAAK;AAAA,MACP;AAAA,IACF,CAAC;AAAA,EACH;AAEA,iBAAe,MAAM,MAA6B;AAChD,QAAI,CAAC,MAAM,GAAG,eAAe,UAAU,MAAM;AAC3C,YAAM,IAAI,MAAM,yBAAyB;AAAA,IAC3C;AAEA,kBAAc,CAAC;AAEf,OAAG;AAAA,MACD,KAAK,UAAU;AAAA,QACb,MAAM;AAAA,QACN;AAAA,MACF,CAAC;AAAA,IACH;AAEA,eAAW,MAAM;AACf,SAAI;AAAA,QACF,KAAK,UAAU;AAAA,UACb,MAAM;AAAA,QACR,CAAC;AAAA,MACH;AAAA,IACF,GAAG,GAAG;AAAA,EACR;AAEA,WAAS,YAAY;AACnB,QAAI,SAAmB;AACvB,QAAI,OAAO,WAAW,YAAa;AAEnC,QAAI;AACF,UAAI,CAAC,cAAc;AACjB,uBAAe,IAAI,aAAa;AAAA,MAClC;AAEA,qBAAe;AAEf,UAAI,aAAa;AACjB,YAAM,WAAyB,CAAC;AAEhC,iBAAW,SAAS,QAAQ;AAC1B,cAAM,eAAe,KAAK,KAAK;AAC/B,cAAM,QAAQ,IAAI,WAAW,aAAa,MAAM;AAChD,iBAAS,IAAI,GAAG,IAAI,aAAa,QAAQ,KAAK;AAC5C,gBAAM,CAAC,IAAI,aAAa,WAAW,CAAC;AAAA,QACtC;AACA,iBAAS,KAAK,KAAK;AACnB,sBAAc,MAAM;AAAA,MACtB;AAEA,YAAM,WAAW,IAAI,WAAW,UAAU;AAC1C,UAAI,SAAS;AACb,iBAAW,SAAS,UAAU;AAC5B,iBAAS,IAAI,OAAO,MAAM;AAC1B,kBAAU,MAAM;AAAA,MAClB;AAEA,YAAM,aAAa,SAAS,SAAS;AACrC,YAAM,cAAc,aAAa,aAAa,GAAG,YAAY,UAAU;AACvE,YAAM,cAAc,YAAY,eAAe,CAAC;AAEhD,YAAM,WAAW,IAAI,SAAS,SAAS,MAAM;AAC7C,eAAS,IAAI,GAAG,IAAI,YAAY,KAAK;AACnC,cAAM,QAAQ,SAAS,SAAS,IAAI,GAAG,IAAI;AAC3C,oBAAY,CAAC,IAAI,QAAQ;AAAA,MAC3B;AAEA,YAAM,SAAS,aAAa,mBAAmB;AAC/C,aAAO,SAAS;AAChB,aAAO,QAAQ,aAAa,WAAW;AACvC,aAAO,UAAU,MAAM,aAAa;AACpC,aAAO,MAAM,CAAC;AACd,oBAAc;AAAA,IAChB,SAAS,KAAK;AACZ,gBAAU,GAAY;AAAA,IACxB;AAAA,EACF;AAEA,WAAS,YAAY;AACnB,QAAI,aAAa;AACf,kBAAY,KAAK;AACjB,oBAAc;AAAA,IAChB;AACA,QAAI,cAAc;AAChB,mBAAa,MAAM;AACnB,qBAAe;AAAA,IACjB;AAAA,EACF;AAEA,WAAS,QAAQ;AACf,QAAI,IAAI;AACN,SAAG,MAAM;AACT,WAAK;AAAA,IACP;AACA,cAAU;AAAA,EACZ;AAEA,SAAO;AAAA,IACL;AAAA,IACA;AAAA,IACA;AAAA,IACA,MAAM;AAAA,IACN,MAAM;AAAA,EACR;AACF;AAEA,IAAO,qBAAQ,CACb,eAC6C;AAC7C,SAAO,CAAC,WAA4B;AAClC,WAAO,gBAAgB,EAAE,GAAG,YAAY,GAAG,OAAO,CAAC;AAAA,EACrD;AACF;","names":[]}
package/dist/index.d.cts CHANGED
@@ -54,6 +54,8 @@ interface TTSClient {
54
54
  speak(text: string): Promise<void>;
55
55
  /** Play audio from chunks */
56
56
  play(): void;
57
+ /** Stop audio playback */
58
+ stop(): void;
57
59
  /** Close connection */
58
60
  close(): void;
59
61
  }
package/dist/index.d.ts CHANGED
@@ -54,6 +54,8 @@ interface TTSClient {
54
54
  speak(text: string): Promise<void>;
55
55
  /** Play audio from chunks */
56
56
  play(): void;
57
+ /** Stop audio playback */
58
+ stop(): void;
57
59
  /** Close connection */
58
60
  close(): void;
59
61
  }
package/dist/index.js CHANGED
@@ -16,6 +16,7 @@ function createTTSClient(config) {
16
16
  let ws = null;
17
17
  let audioChunks = [];
18
18
  let audioContext = null;
19
+ let audioSource = null;
19
20
  async function connect() {
20
21
  return new Promise((resolve, reject) => {
21
22
  let wsUrl = TTS_PATH;
@@ -134,25 +135,34 @@ function createTTSClient(config) {
134
135
  source.connect(audioContext.destination);
135
136
  source.onended = () => onAudioEnd?.();
136
137
  source.start(0);
138
+ audioSource = source;
137
139
  } catch (err) {
138
140
  onError?.(err);
139
141
  }
140
142
  }
141
- function close() {
142
- if (ws) {
143
- ws.close();
144
- ws = null;
143
+ function stopAudio() {
144
+ if (audioSource) {
145
+ audioSource.stop();
146
+ audioSource = null;
145
147
  }
146
148
  if (audioContext) {
147
149
  audioContext.close();
148
150
  audioContext = null;
149
151
  }
150
152
  }
153
+ function close() {
154
+ if (ws) {
155
+ ws.close();
156
+ ws = null;
157
+ }
158
+ stopAudio();
159
+ }
151
160
  return {
152
161
  connect,
153
162
  speak,
154
163
  close,
155
- play: playAudio
164
+ play: playAudio,
165
+ stop: stopAudio
156
166
  };
157
167
  }
158
168
  var tts_client_default = (authConfig) => {
package/dist/index.js.map CHANGED
@@ -1 +1 @@
1
- {"version":3,"sources":["../src/tts-client.ts"],"sourcesContent":["/**\n * TTS Realtime WebSocket Client\n *\n * WebSocket-based real-time text-to-speech synthesis with multiple voice options.\n * Built-in playback only supports PCM format.\n *\n * @example\n * ```typescript\n * const client = createTTSClient({\n * voice: \"Cherry\",\n * autoPlay: true,\n * onReady() {\n * console.log(\"TTS ready\");\n * },\n * onAudioStart() {\n * console.log(\"Playing audio\");\n * },\n * onAudioEnd() {\n * console.log(\"Playback ended\");\n * },\n * });\n *\n * await client.connect();\n * await client.speak(\"Hello, this is a test.\");\n * // client.close();\n * ```\n */\n\nconst TTS_PATH = \"/api/proxy/builtin/platform/qwen-tts/api-ws/v1/realtime\";\n\nexport interface TTSClientConfig {\n /** Get access token for WebSocket authentication */\n getAccessToken?: () => string | null;\n /** Voice name, default 'Cherry' */\n voice?: string;\n /** Auto play audio, default true */\n autoPlay?: boolean;\n /** Audio format, default 'pcm' */\n audioFormat?: \"pcm\" | \"mp3\" | \"wav\" | \"opus\";\n /** Sample rate, default 24000 */\n sampleRate?: number;\n /** Called when connection is ready */\n onReady?: () => void;\n /** Called when audio playback starts */\n onAudioStart?: () => void;\n /** Called when audio playback ends */\n onAudioEnd?: () => void;\n /** Called on each audio chunk received */\n onAudioChunk?: (chunk: string[]) => void;\n /** Called on error */\n onError?: (error: Error) => void;\n}\n\nexport interface TTSClient {\n /** Connect to TTS service */\n connect(): Promise<void>;\n /** Synthesize speech from text */\n speak(text: string): Promise<void>;\n /** Play audio from chunks */\n play(): void;\n /** Close connection */\n close(): void;\n}\n\nfunction createTTSClient(config: TTSClientConfig): TTSClient {\n const {\n getAccessToken,\n voice = \"Cherry\",\n autoPlay = true,\n audioFormat = \"pcm\",\n sampleRate = 24000,\n onReady,\n onAudioStart,\n onAudioEnd,\n onAudioChunk,\n onError,\n } = config;\n\n let ws: WebSocket | null = null;\n let audioChunks: string[] = [];\n let audioContext: AudioContext | null = null;\n\n async function connect(): Promise<void> {\n return new Promise((resolve, reject) => {\n // Build WebSocket URL with optional token parameter\n let wsUrl = TTS_PATH;\n if (getAccessToken) {\n const token = getAccessToken();\n if (token) {\n const separator = wsUrl.includes(\"?\") ? \"&\" : \"?\";\n wsUrl = `${wsUrl}${separator}token=${encodeURIComponent(token)}`;\n }\n }\n\n ws = new WebSocket(wsUrl);\n\n ws.onopen = () => {};\n\n ws.onmessage = (event) => {\n const data = JSON.parse(event.data);\n\n if (data.type === \"session.created\") {\n ws!.send(\n JSON.stringify({\n type: \"session.update\",\n session: {\n mode: \"server_commit\",\n voice,\n response_format: audioFormat,\n sample_rate: sampleRate,\n },\n })\n );\n }\n\n if (data.type === \"session.updated\") {\n onReady?.();\n resolve();\n }\n\n if (data.type === \"response.audio.delta\") {\n audioChunks.push(data.delta);\n onAudioChunk?.(audioChunks);\n }\n\n if (data.type === \"response.audio.done\") {\n onAudioChunk?.(audioChunks);\n if (autoPlay && typeof window !== \"undefined\") {\n playAudio();\n }\n }\n\n if (data.type === \"response.done\") {\n ws!.send(JSON.stringify({ type: \"session.finish\" }));\n }\n\n if (data.type === \"error\") {\n const err = new Error(data.error?.message || \"Unknown error\");\n onError?.(err);\n reject(err);\n }\n };\n\n ws.onerror = () => {\n const err = new Error(\"WebSocket connection error\");\n onError?.(err);\n reject(err);\n };\n\n ws.onclose = () => {\n ws = null;\n };\n });\n }\n\n async function speak(text: string): Promise<void> {\n if (!ws || ws.readyState !== WebSocket.OPEN) {\n throw new Error(\"WebSocket not connected\");\n }\n\n audioChunks = [];\n\n ws.send(\n JSON.stringify({\n type: \"input_text_buffer.append\",\n text,\n })\n );\n\n setTimeout(() => {\n ws!.send(\n JSON.stringify({\n type: \"input_text_buffer.commit\",\n })\n );\n }, 100);\n }\n\n function playAudio() {\n let chunks: string[] = audioChunks;\n if (typeof window === \"undefined\") return;\n\n try {\n if (!audioContext) {\n audioContext = new AudioContext();\n }\n\n onAudioStart?.();\n\n let totalBytes = 0;\n const allBytes: Uint8Array[] = [];\n\n for (const chunk of chunks) {\n const binaryString = atob(chunk);\n const bytes = new Uint8Array(binaryString.length);\n for (let i = 0; i < binaryString.length; i++) {\n bytes[i] = binaryString.charCodeAt(i);\n }\n allBytes.push(bytes);\n totalBytes += bytes.length;\n }\n\n const combined = new Uint8Array(totalBytes);\n let offset = 0;\n for (const bytes of allBytes) {\n combined.set(bytes, offset);\n offset += bytes.length;\n }\n\n const numSamples = combined.length / 2;\n const audioBuffer = audioContext.createBuffer(1, numSamples, sampleRate);\n const channelData = audioBuffer.getChannelData(0);\n\n const dataView = new DataView(combined.buffer);\n for (let i = 0; i < numSamples; i++) {\n const int16 = dataView.getInt16(i * 2, true);\n channelData[i] = int16 / 32768.0;\n }\n\n const source = audioContext.createBufferSource();\n source.buffer = audioBuffer;\n source.connect(audioContext.destination);\n source.onended = () => onAudioEnd?.();\n source.start(0);\n } catch (err) {\n onError?.(err as Error);\n }\n }\n\n function close() {\n if (ws) {\n ws.close();\n ws = null;\n }\n if (audioContext) {\n audioContext.close();\n audioContext = null;\n }\n }\n\n return {\n connect,\n speak,\n close,\n play: playAudio,\n };\n}\n\nexport default (\n authConfig: Pick<TTSClientConfig, \"getAccessToken\">\n): ((config: TTSClientConfig) => TTSClient) => {\n return (config: TTSClientConfig) => {\n return createTTSClient({ ...authConfig, ...config });\n };\n};\n"],"mappings":";AA4BA,IAAM,WAAW;AAoCjB,SAAS,gBAAgB,QAAoC;AAC3D,QAAM;AAAA,IACJ;AAAA,IACA,QAAQ;AAAA,IACR,WAAW;AAAA,IACX,cAAc;AAAA,IACd,aAAa;AAAA,IACb;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF,IAAI;AAEJ,MAAI,KAAuB;AAC3B,MAAI,cAAwB,CAAC;AAC7B,MAAI,eAAoC;AAExC,iBAAe,UAAyB;AACtC,WAAO,IAAI,QAAQ,CAAC,SAAS,WAAW;AAEtC,UAAI,QAAQ;AACZ,UAAI,gBAAgB;AAClB,cAAM,QAAQ,eAAe;AAC7B,YAAI,OAAO;AACT,gBAAM,YAAY,MAAM,SAAS,GAAG,IAAI,MAAM;AAC9C,kBAAQ,GAAG,KAAK,GAAG,SAAS,SAAS,mBAAmB,KAAK,CAAC;AAAA,QAChE;AAAA,MACF;AAEA,WAAK,IAAI,UAAU,KAAK;AAExB,SAAG,SAAS,MAAM;AAAA,MAAC;AAEnB,SAAG,YAAY,CAAC,UAAU;AACxB,cAAM,OAAO,KAAK,MAAM,MAAM,IAAI;AAElC,YAAI,KAAK,SAAS,mBAAmB;AACnC,aAAI;AAAA,YACF,KAAK,UAAU;AAAA,cACb,MAAM;AAAA,cACN,SAAS;AAAA,gBACP,MAAM;AAAA,gBACN;AAAA,gBACA,iBAAiB;AAAA,gBACjB,aAAa;AAAA,cACf;AAAA,YACF,CAAC;AAAA,UACH;AAAA,QACF;AAEA,YAAI,KAAK,SAAS,mBAAmB;AACnC,oBAAU;AACV,kBAAQ;AAAA,QACV;AAEA,YAAI,KAAK,SAAS,wBAAwB;AACxC,sBAAY,KAAK,KAAK,KAAK;AAC3B,yBAAe,WAAW;AAAA,QAC5B;AAEA,YAAI,KAAK,SAAS,uBAAuB;AACvC,yBAAe,WAAW;AAC1B,cAAI,YAAY,OAAO,WAAW,aAAa;AAC7C,sBAAU;AAAA,UACZ;AAAA,QACF;AAEA,YAAI,KAAK,SAAS,iBAAiB;AACjC,aAAI,KAAK,KAAK,UAAU,EAAE,MAAM,iBAAiB,CAAC,CAAC;AAAA,QACrD;AAEA,YAAI,KAAK,SAAS,SAAS;AACzB,gBAAM,MAAM,IAAI,MAAM,KAAK,OAAO,WAAW,eAAe;AAC5D,oBAAU,GAAG;AACb,iBAAO,GAAG;AAAA,QACZ;AAAA,MACF;AAEA,SAAG,UAAU,MAAM;AACjB,cAAM,MAAM,IAAI,MAAM,4BAA4B;AAClD,kBAAU,GAAG;AACb,eAAO,GAAG;AAAA,MACZ;AAEA,SAAG,UAAU,MAAM;AACjB,aAAK;AAAA,MACP;AAAA,IACF,CAAC;AAAA,EACH;AAEA,iBAAe,MAAM,MAA6B;AAChD,QAAI,CAAC,MAAM,GAAG,eAAe,UAAU,MAAM;AAC3C,YAAM,IAAI,MAAM,yBAAyB;AAAA,IAC3C;AAEA,kBAAc,CAAC;AAEf,OAAG;AAAA,MACD,KAAK,UAAU;AAAA,QACb,MAAM;AAAA,QACN;AAAA,MACF,CAAC;AAAA,IACH;AAEA,eAAW,MAAM;AACf,SAAI;AAAA,QACF,KAAK,UAAU;AAAA,UACb,MAAM;AAAA,QACR,CAAC;AAAA,MACH;AAAA,IACF,GAAG,GAAG;AAAA,EACR;AAEA,WAAS,YAAY;AACnB,QAAI,SAAmB;AACvB,QAAI,OAAO,WAAW,YAAa;AAEnC,QAAI;AACF,UAAI,CAAC,cAAc;AACjB,uBAAe,IAAI,aAAa;AAAA,MAClC;AAEA,qBAAe;AAEf,UAAI,aAAa;AACjB,YAAM,WAAyB,CAAC;AAEhC,iBAAW,SAAS,QAAQ;AAC1B,cAAM,eAAe,KAAK,KAAK;AAC/B,cAAM,QAAQ,IAAI,WAAW,aAAa,MAAM;AAChD,iBAAS,IAAI,GAAG,IAAI,aAAa,QAAQ,KAAK;AAC5C,gBAAM,CAAC,IAAI,aAAa,WAAW,CAAC;AAAA,QACtC;AACA,iBAAS,KAAK,KAAK;AACnB,sBAAc,MAAM;AAAA,MACtB;AAEA,YAAM,WAAW,IAAI,WAAW,UAAU;AAC1C,UAAI,SAAS;AACb,iBAAW,SAAS,UAAU;AAC5B,iBAAS,IAAI,OAAO,MAAM;AAC1B,kBAAU,MAAM;AAAA,MAClB;AAEA,YAAM,aAAa,SAAS,SAAS;AACrC,YAAM,cAAc,aAAa,aAAa,GAAG,YAAY,UAAU;AACvE,YAAM,cAAc,YAAY,eAAe,CAAC;AAEhD,YAAM,WAAW,IAAI,SAAS,SAAS,MAAM;AAC7C,eAAS,IAAI,GAAG,IAAI,YAAY,KAAK;AACnC,cAAM,QAAQ,SAAS,SAAS,IAAI,GAAG,IAAI;AAC3C,oBAAY,CAAC,IAAI,QAAQ;AAAA,MAC3B;AAEA,YAAM,SAAS,aAAa,mBAAmB;AAC/C,aAAO,SAAS;AAChB,aAAO,QAAQ,aAAa,WAAW;AACvC,aAAO,UAAU,MAAM,aAAa;AACpC,aAAO,MAAM,CAAC;AAAA,IAChB,SAAS,KAAK;AACZ,gBAAU,GAAY;AAAA,IACxB;AAAA,EACF;AAEA,WAAS,QAAQ;AACf,QAAI,IAAI;AACN,SAAG,MAAM;AACT,WAAK;AAAA,IACP;AACA,QAAI,cAAc;AAChB,mBAAa,MAAM;AACnB,qBAAe;AAAA,IACjB;AAAA,EACF;AAEA,SAAO;AAAA,IACL;AAAA,IACA;AAAA,IACA;AAAA,IACA,MAAM;AAAA,EACR;AACF;AAEA,IAAO,qBAAQ,CACb,eAC6C;AAC7C,SAAO,CAAC,WAA4B;AAClC,WAAO,gBAAgB,EAAE,GAAG,YAAY,GAAG,OAAO,CAAC;AAAA,EACrD;AACF;","names":[]}
1
+ {"version":3,"sources":["../src/tts-client.ts"],"sourcesContent":["/**\n * TTS Realtime WebSocket Client\n *\n * WebSocket-based real-time text-to-speech synthesis with multiple voice options.\n * Built-in playback only supports PCM format.\n *\n * @example\n * ```typescript\n * const client = createTTSClient({\n * voice: \"Cherry\",\n * autoPlay: true,\n * onReady() {\n * console.log(\"TTS ready\");\n * },\n * onAudioStart() {\n * console.log(\"Playing audio\");\n * },\n * onAudioEnd() {\n * console.log(\"Playback ended\");\n * },\n * });\n *\n * await client.connect();\n * await client.speak(\"Hello, this is a test.\");\n * // client.close();\n * ```\n */\n\nconst TTS_PATH = \"/api/proxy/builtin/platform/qwen-tts/api-ws/v1/realtime\";\n\nexport interface TTSClientConfig {\n /** Get access token for WebSocket authentication */\n getAccessToken?: () => string | null;\n /** Voice name, default 'Cherry' */\n voice?: string;\n /** Auto play audio, default true */\n autoPlay?: boolean;\n /** Audio format, default 'pcm' */\n audioFormat?: \"pcm\" | \"mp3\" | \"wav\" | \"opus\";\n /** Sample rate, default 24000 */\n sampleRate?: number;\n /** Called when connection is ready */\n onReady?: () => void;\n /** Called when audio playback starts */\n onAudioStart?: () => void;\n /** Called when audio playback ends */\n onAudioEnd?: () => void;\n /** Called on each audio chunk received */\n onAudioChunk?: (chunk: string[]) => void;\n /** Called on error */\n onError?: (error: Error) => void;\n}\n\nexport interface TTSClient {\n /** Connect to TTS service */\n connect(): Promise<void>;\n /** Synthesize speech from text */\n speak(text: string): Promise<void>;\n /** Play audio from chunks */\n play(): void;\n /** Stop audio playback */\n stop(): void;\n /** Close connection */\n close(): void;\n}\n\nfunction createTTSClient(config: TTSClientConfig): TTSClient {\n const {\n getAccessToken,\n voice = \"Cherry\",\n autoPlay = true,\n audioFormat = \"pcm\",\n sampleRate = 24000,\n onReady,\n onAudioStart,\n onAudioEnd,\n onAudioChunk,\n onError,\n } = config;\n\n let ws: WebSocket | null = null;\n let audioChunks: string[] = [];\n let audioContext: AudioContext | null = null;\n let audioSource: AudioBufferSourceNode | null = null;\n\n async function connect(): Promise<void> {\n return new Promise((resolve, reject) => {\n // Build WebSocket URL with optional token parameter\n let wsUrl = TTS_PATH;\n if (getAccessToken) {\n const token = getAccessToken();\n if (token) {\n const separator = wsUrl.includes(\"?\") ? \"&\" : \"?\";\n wsUrl = `${wsUrl}${separator}token=${encodeURIComponent(token)}`;\n }\n }\n\n ws = new WebSocket(wsUrl);\n\n ws.onopen = () => {};\n\n ws.onmessage = (event) => {\n const data = JSON.parse(event.data);\n\n if (data.type === \"session.created\") {\n ws!.send(\n JSON.stringify({\n type: \"session.update\",\n session: {\n mode: \"server_commit\",\n voice,\n response_format: audioFormat,\n sample_rate: sampleRate,\n },\n })\n );\n }\n\n if (data.type === \"session.updated\") {\n onReady?.();\n resolve();\n }\n\n if (data.type === \"response.audio.delta\") {\n audioChunks.push(data.delta);\n onAudioChunk?.(audioChunks);\n }\n\n if (data.type === \"response.audio.done\") {\n onAudioChunk?.(audioChunks);\n if (autoPlay && typeof window !== \"undefined\") {\n playAudio();\n }\n }\n\n if (data.type === \"response.done\") {\n ws!.send(JSON.stringify({ type: \"session.finish\" }));\n }\n\n if (data.type === \"error\") {\n const err = new Error(data.error?.message || \"Unknown error\");\n onError?.(err);\n reject(err);\n }\n };\n\n ws.onerror = () => {\n const err = new Error(\"WebSocket connection error\");\n onError?.(err);\n reject(err);\n };\n\n ws.onclose = () => {\n ws = null;\n };\n });\n }\n\n async function speak(text: string): Promise<void> {\n if (!ws || ws.readyState !== WebSocket.OPEN) {\n throw new Error(\"WebSocket not connected\");\n }\n\n audioChunks = [];\n\n ws.send(\n JSON.stringify({\n type: \"input_text_buffer.append\",\n text,\n })\n );\n\n setTimeout(() => {\n ws!.send(\n JSON.stringify({\n type: \"input_text_buffer.commit\",\n })\n );\n }, 100);\n }\n\n function playAudio() {\n let chunks: string[] = audioChunks;\n if (typeof window === \"undefined\") return;\n\n try {\n if (!audioContext) {\n audioContext = new AudioContext();\n }\n\n onAudioStart?.();\n\n let totalBytes = 0;\n const allBytes: Uint8Array[] = [];\n\n for (const chunk of chunks) {\n const binaryString = atob(chunk);\n const bytes = new Uint8Array(binaryString.length);\n for (let i = 0; i < binaryString.length; i++) {\n bytes[i] = binaryString.charCodeAt(i);\n }\n allBytes.push(bytes);\n totalBytes += bytes.length;\n }\n\n const combined = new Uint8Array(totalBytes);\n let offset = 0;\n for (const bytes of allBytes) {\n combined.set(bytes, offset);\n offset += bytes.length;\n }\n\n const numSamples = combined.length / 2;\n const audioBuffer = audioContext.createBuffer(1, numSamples, sampleRate);\n const channelData = audioBuffer.getChannelData(0);\n\n const dataView = new DataView(combined.buffer);\n for (let i = 0; i < numSamples; i++) {\n const int16 = dataView.getInt16(i * 2, true);\n channelData[i] = int16 / 32768.0;\n }\n\n const source = audioContext.createBufferSource();\n source.buffer = audioBuffer;\n source.connect(audioContext.destination);\n source.onended = () => onAudioEnd?.();\n source.start(0);\n audioSource = source;\n } catch (err) {\n onError?.(err as Error);\n }\n }\n\n function stopAudio() {\n if (audioSource) {\n audioSource.stop();\n audioSource = null;\n }\n if (audioContext) {\n audioContext.close();\n audioContext = null;\n }\n }\n\n function close() {\n if (ws) {\n ws.close();\n ws = null;\n }\n stopAudio();\n }\n\n return {\n connect,\n speak,\n close,\n play: playAudio,\n stop: stopAudio,\n };\n}\n\nexport default (\n authConfig: Pick<TTSClientConfig, \"getAccessToken\">\n): ((config: TTSClientConfig) => TTSClient) => {\n return (config: TTSClientConfig) => {\n return createTTSClient({ ...authConfig, ...config });\n };\n};\n"],"mappings":";AA4BA,IAAM,WAAW;AAsCjB,SAAS,gBAAgB,QAAoC;AAC3D,QAAM;AAAA,IACJ;AAAA,IACA,QAAQ;AAAA,IACR,WAAW;AAAA,IACX,cAAc;AAAA,IACd,aAAa;AAAA,IACb;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF,IAAI;AAEJ,MAAI,KAAuB;AAC3B,MAAI,cAAwB,CAAC;AAC7B,MAAI,eAAoC;AACxC,MAAI,cAA4C;AAEhD,iBAAe,UAAyB;AACtC,WAAO,IAAI,QAAQ,CAAC,SAAS,WAAW;AAEtC,UAAI,QAAQ;AACZ,UAAI,gBAAgB;AAClB,cAAM,QAAQ,eAAe;AAC7B,YAAI,OAAO;AACT,gBAAM,YAAY,MAAM,SAAS,GAAG,IAAI,MAAM;AAC9C,kBAAQ,GAAG,KAAK,GAAG,SAAS,SAAS,mBAAmB,KAAK,CAAC;AAAA,QAChE;AAAA,MACF;AAEA,WAAK,IAAI,UAAU,KAAK;AAExB,SAAG,SAAS,MAAM;AAAA,MAAC;AAEnB,SAAG,YAAY,CAAC,UAAU;AACxB,cAAM,OAAO,KAAK,MAAM,MAAM,IAAI;AAElC,YAAI,KAAK,SAAS,mBAAmB;AACnC,aAAI;AAAA,YACF,KAAK,UAAU;AAAA,cACb,MAAM;AAAA,cACN,SAAS;AAAA,gBACP,MAAM;AAAA,gBACN;AAAA,gBACA,iBAAiB;AAAA,gBACjB,aAAa;AAAA,cACf;AAAA,YACF,CAAC;AAAA,UACH;AAAA,QACF;AAEA,YAAI,KAAK,SAAS,mBAAmB;AACnC,oBAAU;AACV,kBAAQ;AAAA,QACV;AAEA,YAAI,KAAK,SAAS,wBAAwB;AACxC,sBAAY,KAAK,KAAK,KAAK;AAC3B,yBAAe,WAAW;AAAA,QAC5B;AAEA,YAAI,KAAK,SAAS,uBAAuB;AACvC,yBAAe,WAAW;AAC1B,cAAI,YAAY,OAAO,WAAW,aAAa;AAC7C,sBAAU;AAAA,UACZ;AAAA,QACF;AAEA,YAAI,KAAK,SAAS,iBAAiB;AACjC,aAAI,KAAK,KAAK,UAAU,EAAE,MAAM,iBAAiB,CAAC,CAAC;AAAA,QACrD;AAEA,YAAI,KAAK,SAAS,SAAS;AACzB,gBAAM,MAAM,IAAI,MAAM,KAAK,OAAO,WAAW,eAAe;AAC5D,oBAAU,GAAG;AACb,iBAAO,GAAG;AAAA,QACZ;AAAA,MACF;AAEA,SAAG,UAAU,MAAM;AACjB,cAAM,MAAM,IAAI,MAAM,4BAA4B;AAClD,kBAAU,GAAG;AACb,eAAO,GAAG;AAAA,MACZ;AAEA,SAAG,UAAU,MAAM;AACjB,aAAK;AAAA,MACP;AAAA,IACF,CAAC;AAAA,EACH;AAEA,iBAAe,MAAM,MAA6B;AAChD,QAAI,CAAC,MAAM,GAAG,eAAe,UAAU,MAAM;AAC3C,YAAM,IAAI,MAAM,yBAAyB;AAAA,IAC3C;AAEA,kBAAc,CAAC;AAEf,OAAG;AAAA,MACD,KAAK,UAAU;AAAA,QACb,MAAM;AAAA,QACN;AAAA,MACF,CAAC;AAAA,IACH;AAEA,eAAW,MAAM;AACf,SAAI;AAAA,QACF,KAAK,UAAU;AAAA,UACb,MAAM;AAAA,QACR,CAAC;AAAA,MACH;AAAA,IACF,GAAG,GAAG;AAAA,EACR;AAEA,WAAS,YAAY;AACnB,QAAI,SAAmB;AACvB,QAAI,OAAO,WAAW,YAAa;AAEnC,QAAI;AACF,UAAI,CAAC,cAAc;AACjB,uBAAe,IAAI,aAAa;AAAA,MAClC;AAEA,qBAAe;AAEf,UAAI,aAAa;AACjB,YAAM,WAAyB,CAAC;AAEhC,iBAAW,SAAS,QAAQ;AAC1B,cAAM,eAAe,KAAK,KAAK;AAC/B,cAAM,QAAQ,IAAI,WAAW,aAAa,MAAM;AAChD,iBAAS,IAAI,GAAG,IAAI,aAAa,QAAQ,KAAK;AAC5C,gBAAM,CAAC,IAAI,aAAa,WAAW,CAAC;AAAA,QACtC;AACA,iBAAS,KAAK,KAAK;AACnB,sBAAc,MAAM;AAAA,MACtB;AAEA,YAAM,WAAW,IAAI,WAAW,UAAU;AAC1C,UAAI,SAAS;AACb,iBAAW,SAAS,UAAU;AAC5B,iBAAS,IAAI,OAAO,MAAM;AAC1B,kBAAU,MAAM;AAAA,MAClB;AAEA,YAAM,aAAa,SAAS,SAAS;AACrC,YAAM,cAAc,aAAa,aAAa,GAAG,YAAY,UAAU;AACvE,YAAM,cAAc,YAAY,eAAe,CAAC;AAEhD,YAAM,WAAW,IAAI,SAAS,SAAS,MAAM;AAC7C,eAAS,IAAI,GAAG,IAAI,YAAY,KAAK;AACnC,cAAM,QAAQ,SAAS,SAAS,IAAI,GAAG,IAAI;AAC3C,oBAAY,CAAC,IAAI,QAAQ;AAAA,MAC3B;AAEA,YAAM,SAAS,aAAa,mBAAmB;AAC/C,aAAO,SAAS;AAChB,aAAO,QAAQ,aAAa,WAAW;AACvC,aAAO,UAAU,MAAM,aAAa;AACpC,aAAO,MAAM,CAAC;AACd,oBAAc;AAAA,IAChB,SAAS,KAAK;AACZ,gBAAU,GAAY;AAAA,IACxB;AAAA,EACF;AAEA,WAAS,YAAY;AACnB,QAAI,aAAa;AACf,kBAAY,KAAK;AACjB,oBAAc;AAAA,IAChB;AACA,QAAI,cAAc;AAChB,mBAAa,MAAM;AACnB,qBAAe;AAAA,IACjB;AAAA,EACF;AAEA,WAAS,QAAQ;AACf,QAAI,IAAI;AACN,SAAG,MAAM;AACT,WAAK;AAAA,IACP;AACA,cAAU;AAAA,EACZ;AAEA,SAAO;AAAA,IACL;AAAA,IACA;AAAA,IACA;AAAA,IACA,MAAM;AAAA,IACN,MAAM;AAAA,EACR;AACF;AAEA,IAAO,qBAAQ,CACb,eAC6C;AAC7C,SAAO,CAAC,WAA4B;AAClC,WAAO,gBAAgB,EAAE,GAAG,YAAY,GAAG,OAAO,CAAC;AAAA,EACrD;AACF;","names":[]}
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@amaster.ai/tts-client",
3
- "version": "1.1.0-beta.51",
3
+ "version": "1.1.0-beta.53",
4
4
  "description": "Qwen TTS Realtime WebSocket client with audio playback",
5
5
  "type": "module",
6
6
  "main": "./dist/index.cjs",