@livekit/agents-plugin-openai 1.0.22 → 1.0.23
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/stt.cjs +16 -7
- package/dist/stt.cjs.map +1 -1
- package/dist/stt.d.cts +2 -0
- package/dist/stt.d.ts +2 -0
- package/dist/stt.d.ts.map +1 -1
- package/dist/stt.js +16 -7
- package/dist/stt.js.map +1 -1
- package/package.json +5 -5
- package/src/stt.ts +18 -7
package/dist/stt.cjs
CHANGED
|
@@ -33,6 +33,7 @@ class STT extends import_agents.stt.STT {
|
|
|
33
33
|
#opts;
|
|
34
34
|
#client;
|
|
35
35
|
label = "openai.STT";
|
|
36
|
+
abortController = new AbortController();
|
|
36
37
|
/**
|
|
37
38
|
* Create a new instance of OpenAI STT.
|
|
38
39
|
*
|
|
@@ -120,13 +121,18 @@ class STT extends import_agents.stt.STT {
|
|
|
120
121
|
const config = this.#sanitizeOptions(language);
|
|
121
122
|
buffer = (0, import_agents.mergeFrames)(buffer);
|
|
122
123
|
const file = new File([this.#createWav(buffer)], "audio.wav", { type: "audio/wav" });
|
|
123
|
-
const resp = await this.#client.audio.transcriptions.create(
|
|
124
|
-
|
|
125
|
-
|
|
126
|
-
|
|
127
|
-
|
|
128
|
-
|
|
129
|
-
|
|
124
|
+
const resp = await this.#client.audio.transcriptions.create(
|
|
125
|
+
{
|
|
126
|
+
file,
|
|
127
|
+
model: this.#opts.model,
|
|
128
|
+
language: config.language,
|
|
129
|
+
prompt: config.prompt,
|
|
130
|
+
response_format: "json"
|
|
131
|
+
},
|
|
132
|
+
{
|
|
133
|
+
signal: this.abortController.signal
|
|
134
|
+
}
|
|
135
|
+
);
|
|
130
136
|
return {
|
|
131
137
|
type: import_agents.stt.SpeechEventType.FINAL_TRANSCRIPT,
|
|
132
138
|
alternatives: [
|
|
@@ -144,6 +150,9 @@ class STT extends import_agents.stt.STT {
|
|
|
144
150
|
stream() {
|
|
145
151
|
throw new Error("Streaming is not supported on OpenAI STT");
|
|
146
152
|
}
|
|
153
|
+
async close() {
|
|
154
|
+
this.abortController.abort();
|
|
155
|
+
}
|
|
147
156
|
}
|
|
148
157
|
// Annotate the CommonJS export names for ESM import in node:
|
|
149
158
|
0 && (module.exports = {
|
package/dist/stt.cjs.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["../src/stt.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { type AudioBuffer, mergeFrames, stt } from '@livekit/agents';\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport { OpenAI } from 'openai';\nimport type { GroqAudioModels, WhisperModels } from './models.js';\n\nexport interface STTOptions {\n apiKey?: string;\n language: string;\n prompt?: string;\n detectLanguage: boolean;\n model: WhisperModels | string;\n baseURL?: string;\n client?: OpenAI;\n}\n\nconst defaultSTTOptions: STTOptions = {\n apiKey: process.env.OPENAI_API_KEY,\n language: 'en',\n detectLanguage: false,\n model: 'whisper-1',\n};\n\nexport class STT extends stt.STT {\n #opts: STTOptions;\n #client: OpenAI;\n label = 'openai.STT';\n\n /**\n * Create a new instance of OpenAI STT.\n *\n * @remarks\n * `apiKey` must be set to your OpenAI API key, either using the argument or by setting the\n * `OPENAI_API_KEY` environment variable.\n */\n constructor(opts: Partial<STTOptions> = defaultSTTOptions) {\n super({ streaming: false, interimResults: false });\n\n this.#opts = { ...defaultSTTOptions, ...opts };\n if (this.#opts.apiKey === undefined) {\n throw new Error('OpenAI API key is required, whether as an argument or as $OPENAI_API_KEY');\n }\n\n this.#client =\n this.#opts.client ||\n new OpenAI({\n baseURL: opts.baseURL,\n apiKey: opts.apiKey,\n });\n }\n\n /**\n * Create a new instance of Groq STT.\n *\n * @remarks\n * `apiKey` must be set to your Groq API key, either using the argument or by setting the\n * `GROQ_API_KEY` environment variable.\n */\n static withGroq(\n opts: Partial<{\n model: string | GroqAudioModels;\n apiKey?: string;\n baseURL?: string;\n client: OpenAI;\n language: string;\n detectLanguage: boolean;\n }> = {},\n ): STT {\n opts.apiKey = opts.apiKey || process.env.GROQ_API_KEY;\n if (opts.apiKey === undefined) {\n throw new Error('Groq API key is required, whether as an argument or as $GROQ_API_KEY');\n }\n\n return new STT({\n model: 'whisper-large-v3-turbo',\n baseURL: 'https://api.groq.com/openai/v1',\n ...opts,\n });\n }\n\n /**\n * Create a new instance of OVHcloud AI Endpoints STT.\n *\n * @remarks\n * `apiKey` must be set to your OVHcloud AI Endpoints API key, either using the argument or by setting the\n * `OVHCLOUD_API_KEY` environment variable.\n */\n static withOVHcloud(\n opts: Partial<{\n model: string;\n apiKey?: string;\n baseURL?: string;\n client: OpenAI;\n language: string;\n detectLanguage: boolean;\n }> = {},\n ): STT {\n opts.apiKey = opts.apiKey || process.env.OVHCLOUD_API_KEY;\n if (opts.apiKey === undefined) {\n throw new Error(\n 'OVHcloud AI Endpoints API key is required, whether as an argument or as $OVHCLOUD_API_KEY',\n );\n }\n\n return new STT({\n model: 'whisper-large-v3-turbo',\n baseURL: 'https://oai.endpoints.kepler.ai.cloud.ovh.net/v1',\n ...opts,\n });\n }\n\n #sanitizeOptions(language?: string): STTOptions {\n if (language) {\n return { ...this.#opts, language };\n } else {\n return this.#opts;\n }\n }\n\n #createWav(frame: AudioFrame): Buffer {\n const bitsPerSample = 16;\n const byteRate = (frame.sampleRate * frame.channels * bitsPerSample) / 8;\n const blockAlign = (frame.channels * bitsPerSample) / 8;\n\n const header = Buffer.alloc(44);\n header.write('RIFF', 0);\n header.writeUInt32LE(36 + frame.data.byteLength, 4);\n header.write('WAVE', 8);\n header.write('fmt ', 12);\n header.writeUInt32LE(16, 16);\n header.writeUInt16LE(1, 20);\n header.writeUInt16LE(frame.channels, 22);\n header.writeUInt32LE(frame.sampleRate, 24);\n header.writeUInt32LE(byteRate, 28);\n header.writeUInt16LE(blockAlign, 32);\n header.writeUInt16LE(16, 34);\n header.write('data', 36);\n header.writeUInt32LE(frame.data.byteLength, 40);\n return Buffer.concat([header, Buffer.from(frame.data.buffer)]);\n }\n\n async _recognize(buffer: AudioBuffer, language?: string): Promise<stt.SpeechEvent> {\n const config = this.#sanitizeOptions(language);\n buffer = mergeFrames(buffer);\n const file = new File([this.#createWav(buffer)], 'audio.wav', { type: 'audio/wav' });\n const resp = await this.#client.audio.transcriptions.create({\n
|
|
1
|
+
{"version":3,"sources":["../src/stt.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { type AudioBuffer, mergeFrames, stt } from '@livekit/agents';\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport { OpenAI } from 'openai';\nimport type { GroqAudioModels, WhisperModels } from './models.js';\n\nexport interface STTOptions {\n apiKey?: string;\n language: string;\n prompt?: string;\n detectLanguage: boolean;\n model: WhisperModels | string;\n baseURL?: string;\n client?: OpenAI;\n}\n\nconst defaultSTTOptions: STTOptions = {\n apiKey: process.env.OPENAI_API_KEY,\n language: 'en',\n detectLanguage: false,\n model: 'whisper-1',\n};\n\nexport class STT extends stt.STT {\n #opts: STTOptions;\n #client: OpenAI;\n label = 'openai.STT';\n private abortController = new AbortController();\n\n /**\n * Create a new instance of OpenAI STT.\n *\n * @remarks\n * `apiKey` must be set to your OpenAI API key, either using the argument or by setting the\n * `OPENAI_API_KEY` environment variable.\n */\n constructor(opts: Partial<STTOptions> = defaultSTTOptions) {\n super({ streaming: false, interimResults: false });\n\n this.#opts = { ...defaultSTTOptions, ...opts };\n if (this.#opts.apiKey === undefined) {\n throw new Error('OpenAI API key is required, whether as an argument or as $OPENAI_API_KEY');\n }\n\n this.#client =\n this.#opts.client ||\n new OpenAI({\n baseURL: opts.baseURL,\n apiKey: opts.apiKey,\n });\n }\n\n /**\n * Create a new instance of Groq STT.\n *\n * @remarks\n * `apiKey` must be set to your Groq API key, either using the argument or by setting the\n * `GROQ_API_KEY` environment variable.\n */\n static withGroq(\n opts: Partial<{\n model: string | GroqAudioModels;\n apiKey?: string;\n baseURL?: string;\n client: OpenAI;\n language: string;\n detectLanguage: boolean;\n }> = {},\n ): STT {\n opts.apiKey = opts.apiKey || process.env.GROQ_API_KEY;\n if (opts.apiKey === undefined) {\n throw new Error('Groq API key is required, whether as an argument or as $GROQ_API_KEY');\n }\n\n return new STT({\n model: 'whisper-large-v3-turbo',\n baseURL: 'https://api.groq.com/openai/v1',\n ...opts,\n });\n }\n\n /**\n * Create a new instance of OVHcloud AI Endpoints STT.\n *\n * @remarks\n * `apiKey` must be set to your OVHcloud AI Endpoints API key, either using the argument or by setting the\n * `OVHCLOUD_API_KEY` environment variable.\n */\n static withOVHcloud(\n opts: Partial<{\n model: string;\n apiKey?: string;\n baseURL?: string;\n client: OpenAI;\n language: string;\n detectLanguage: boolean;\n }> = {},\n ): STT {\n opts.apiKey = opts.apiKey || process.env.OVHCLOUD_API_KEY;\n if (opts.apiKey === undefined) {\n throw new Error(\n 'OVHcloud AI Endpoints API key is required, whether as an argument or as $OVHCLOUD_API_KEY',\n );\n }\n\n return new STT({\n model: 'whisper-large-v3-turbo',\n baseURL: 'https://oai.endpoints.kepler.ai.cloud.ovh.net/v1',\n ...opts,\n });\n }\n\n #sanitizeOptions(language?: string): STTOptions {\n if (language) {\n return { ...this.#opts, language };\n } else {\n return this.#opts;\n }\n }\n\n #createWav(frame: AudioFrame): Buffer {\n const bitsPerSample = 16;\n const byteRate = (frame.sampleRate * frame.channels * bitsPerSample) / 8;\n const blockAlign = (frame.channels * bitsPerSample) / 8;\n\n const header = Buffer.alloc(44);\n header.write('RIFF', 0);\n header.writeUInt32LE(36 + frame.data.byteLength, 4);\n header.write('WAVE', 8);\n header.write('fmt ', 12);\n header.writeUInt32LE(16, 16);\n header.writeUInt16LE(1, 20);\n header.writeUInt16LE(frame.channels, 22);\n header.writeUInt32LE(frame.sampleRate, 24);\n header.writeUInt32LE(byteRate, 28);\n header.writeUInt16LE(blockAlign, 32);\n header.writeUInt16LE(16, 34);\n header.write('data', 36);\n header.writeUInt32LE(frame.data.byteLength, 40);\n return Buffer.concat([header, Buffer.from(frame.data.buffer)]);\n }\n\n async _recognize(buffer: AudioBuffer, language?: string): Promise<stt.SpeechEvent> {\n const config = this.#sanitizeOptions(language);\n buffer = mergeFrames(buffer);\n const file = new File([this.#createWav(buffer)], 'audio.wav', { type: 'audio/wav' });\n\n const resp = await this.#client.audio.transcriptions.create(\n {\n file,\n model: this.#opts.model,\n language: config.language,\n prompt: config.prompt,\n response_format: 'json',\n },\n {\n signal: this.abortController.signal,\n },\n );\n\n return {\n type: stt.SpeechEventType.FINAL_TRANSCRIPT,\n alternatives: [\n {\n text: resp.text || '',\n language: language || '',\n startTime: 0,\n endTime: 0,\n confidence: 0,\n },\n ],\n };\n }\n\n /** This method throws an error; streaming is unsupported on OpenAI STT. */\n stream(): stt.SpeechStream {\n throw new Error('Streaming is not supported on OpenAI STT');\n }\n\n async close(): Promise<void> {\n this.abortController.abort();\n }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAGA,oBAAmD;AAEnD,oBAAuB;AAavB,MAAM,oBAAgC;AAAA,EACpC,QAAQ,QAAQ,IAAI;AAAA,EACpB,UAAU;AAAA,EACV,gBAAgB;AAAA,EAChB,OAAO;AACT;AAEO,MAAM,YAAY,kBAAI,IAAI;AAAA,EAC/B;AAAA,EACA;AAAA,EACA,QAAQ;AAAA,EACA,kBAAkB,IAAI,gBAAgB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAS9C,YAAY,OAA4B,mBAAmB;AACzD,UAAM,EAAE,WAAW,OAAO,gBAAgB,MAAM,CAAC;AAEjD,SAAK,QAAQ,EAAE,GAAG,mBAAmB,GAAG,KAAK;AAC7C,QAAI,KAAK,MAAM,WAAW,QAAW;AACnC,YAAM,IAAI,MAAM,0EAA0E;AAAA,IAC5F;AAEA,SAAK,UACH,KAAK,MAAM,UACX,IAAI,qBAAO;AAAA,MACT,SAAS,KAAK;AAAA,MACd,QAAQ,KAAK;AAAA,IACf,CAAC;AAAA,EACL;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EASA,OAAO,SACL,OAOK,CAAC,GACD;AACL,SAAK,SAAS,KAAK,UAAU,QAAQ,IAAI;AACzC,QAAI,KAAK,WAAW,QAAW;AAC7B,YAAM,IAAI,MAAM,sEAAsE;AAAA,IACxF;AAEA,WAAO,IAAI,IAAI;AAAA,MACb,OAAO;AAAA,MACP,SAAS;AAAA,MACT,GAAG;AAAA,IACL,CAAC;AAAA,EACH;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EASA,OAAO,aACL,OAOK,CAAC,GACD;AACL,SAAK,SAAS,KAAK,UAAU,QAAQ,IAAI;AACzC,QAAI,KAAK,WAAW,QAAW;AAC7B,YAAM,IAAI;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAEA,WAAO,IAAI,IAAI;AAAA,MACb,OAAO;AAAA,MACP,SAAS;AAAA,MACT,GAAG;AAAA,IACL,CAAC;AAAA,EACH;AAAA,EAEA,iBAAiB,UAA+B;AAC9C,QAAI,UAAU;AACZ,aAAO,EAAE,GAAG,KAAK,OAAO,SAAS;AAAA,IACnC,OAAO;AACL,aAAO,KAAK;AAAA,IACd;AAAA,EACF;AAAA,EAEA,WAAW,OAA2B;AACpC,UAAM,gBAAgB;AACtB,UAAM,WAAY,MAAM,aAAa,MAAM,WAAW,gBAAiB;AACvE,UAAM,aAAc,MAAM,WAAW,gBAAiB;AAEtD,UAAM,SAAS,OAAO,MAAM,EAAE;AAC9B,WAAO,MAAM,QAAQ,CAAC;AACtB,WAAO,cAAc,KAAK,MAAM,KAAK,YAAY,CAAC;AAClD,WAAO,MAAM,QAAQ,CAAC;AACtB,WAAO,MAAM,QAAQ,EAAE;AACvB,WAAO,cAAc,IAAI,EAAE;AAC3B,WAAO,cAAc,GAAG,EAAE;AAC1B,WAAO,cAAc,MAAM,UAAU,EAAE;AACvC,WAAO,cAAc,MAAM,YAAY,EAAE;AACzC,WAAO,cAAc,UAAU,EAAE;AACjC,WAAO,cAAc,YAAY,EAAE;AACnC,WAAO,cAAc,IAAI,EAAE;AAC3B,WAAO,MAAM,QAAQ,EAAE;AACvB,WAAO,cAAc,MAAM,KAAK,YAAY,EAAE;AAC9C,WAAO,OAAO,OAAO,CAAC,QAAQ,OAAO,KAAK,MAAM,KAAK,MAAM,CAAC,CAAC;AAAA,EAC/D;AAAA,EAEA,MAAM,WAAW,QAAqB,UAA6C;AACjF,UAAM,SAAS,KAAK,iBAAiB,QAAQ;AAC7C,iBAAS,2BAAY,MAAM;AAC3B,UAAM,OAAO,IAAI,KAAK,CAAC,KAAK,WAAW,MAAM,CAAC,GAAG,aAAa,EAAE,MAAM,YAAY,CAAC;AAEnF,UAAM,OAAO,MAAM,KAAK,QAAQ,MAAM,eAAe;AAAA,MACnD;AAAA,QACE;AAAA,QACA,OAAO,KAAK,MAAM;AAAA,QAClB,UAAU,OAAO;AAAA,QACjB,QAAQ,OAAO;AAAA,QACf,iBAAiB;AAAA,MACnB;AAAA,MACA;AAAA,QACE,QAAQ,KAAK,gBAAgB;AAAA,MAC/B;AAAA,IACF;AAEA,WAAO;AAAA,MACL,MAAM,kBAAI,gBAAgB;AAAA,MAC1B,cAAc;AAAA,QACZ;AAAA,UACE,MAAM,KAAK,QAAQ;AAAA,UACnB,UAAU,YAAY;AAAA,UACtB,WAAW;AAAA,UACX,SAAS;AAAA,UACT,YAAY;AAAA,QACd;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAAA;AAAA,EAGA,SAA2B;AACzB,UAAM,IAAI,MAAM,0CAA0C;AAAA,EAC5D;AAAA,EAEA,MAAM,QAAuB;AAC3B,SAAK,gBAAgB,MAAM;AAAA,EAC7B;AACF;","names":[]}
|
package/dist/stt.d.cts
CHANGED
|
@@ -13,6 +13,7 @@ export interface STTOptions {
|
|
|
13
13
|
export declare class STT extends stt.STT {
|
|
14
14
|
#private;
|
|
15
15
|
label: string;
|
|
16
|
+
private abortController;
|
|
16
17
|
/**
|
|
17
18
|
* Create a new instance of OpenAI STT.
|
|
18
19
|
*
|
|
@@ -54,5 +55,6 @@ export declare class STT extends stt.STT {
|
|
|
54
55
|
_recognize(buffer: AudioBuffer, language?: string): Promise<stt.SpeechEvent>;
|
|
55
56
|
/** This method throws an error; streaming is unsupported on OpenAI STT. */
|
|
56
57
|
stream(): stt.SpeechStream;
|
|
58
|
+
close(): Promise<void>;
|
|
57
59
|
}
|
|
58
60
|
//# sourceMappingURL=stt.d.ts.map
|
package/dist/stt.d.ts
CHANGED
|
@@ -13,6 +13,7 @@ export interface STTOptions {
|
|
|
13
13
|
export declare class STT extends stt.STT {
|
|
14
14
|
#private;
|
|
15
15
|
label: string;
|
|
16
|
+
private abortController;
|
|
16
17
|
/**
|
|
17
18
|
* Create a new instance of OpenAI STT.
|
|
18
19
|
*
|
|
@@ -54,5 +55,6 @@ export declare class STT extends stt.STT {
|
|
|
54
55
|
_recognize(buffer: AudioBuffer, language?: string): Promise<stt.SpeechEvent>;
|
|
55
56
|
/** This method throws an error; streaming is unsupported on OpenAI STT. */
|
|
56
57
|
stream(): stt.SpeechStream;
|
|
58
|
+
close(): Promise<void>;
|
|
57
59
|
}
|
|
58
60
|
//# sourceMappingURL=stt.d.ts.map
|
package/dist/stt.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"stt.d.ts","sourceRoot":"","sources":["../src/stt.ts"],"names":[],"mappings":"AAGA,OAAO,EAAE,KAAK,WAAW,EAAe,GAAG,EAAE,MAAM,iBAAiB,CAAC;AAErE,OAAO,EAAE,MAAM,EAAE,MAAM,QAAQ,CAAC;AAChC,OAAO,KAAK,EAAE,eAAe,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AAElE,MAAM,WAAW,UAAU;IACzB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,QAAQ,EAAE,MAAM,CAAC;IACjB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,cAAc,EAAE,OAAO,CAAC;IACxB,KAAK,EAAE,aAAa,GAAG,MAAM,CAAC;IAC9B,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AASD,qBAAa,GAAI,SAAQ,GAAG,CAAC,GAAG;;IAG9B,KAAK,SAAgB;
|
|
1
|
+
{"version":3,"file":"stt.d.ts","sourceRoot":"","sources":["../src/stt.ts"],"names":[],"mappings":"AAGA,OAAO,EAAE,KAAK,WAAW,EAAe,GAAG,EAAE,MAAM,iBAAiB,CAAC;AAErE,OAAO,EAAE,MAAM,EAAE,MAAM,QAAQ,CAAC;AAChC,OAAO,KAAK,EAAE,eAAe,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AAElE,MAAM,WAAW,UAAU;IACzB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,QAAQ,EAAE,MAAM,CAAC;IACjB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,cAAc,EAAE,OAAO,CAAC;IACxB,KAAK,EAAE,aAAa,GAAG,MAAM,CAAC;IAC9B,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AASD,qBAAa,GAAI,SAAQ,GAAG,CAAC,GAAG;;IAG9B,KAAK,SAAgB;IACrB,OAAO,CAAC,eAAe,CAAyB;IAEhD;;;;;;OAMG;gBACS,IAAI,GAAE,OAAO,CAAC,UAAU,CAAqB;IAgBzD;;;;;;OAMG;IACH,MAAM,CAAC,QAAQ,CACb,IAAI,GAAE,OAAO,CAAC;QACZ,KAAK,EAAE,MAAM,GAAG,eAAe,CAAC;QAChC,MAAM,CAAC,EAAE,MAAM,CAAC;QAChB,OAAO,CAAC,EAAE,MAAM,CAAC;QACjB,MAAM,EAAE,MAAM,CAAC;QACf,QAAQ,EAAE,MAAM,CAAC;QACjB,cAAc,EAAE,OAAO,CAAC;KACzB,CAAM,GACN,GAAG;IAaN;;;;;;OAMG;IACH,MAAM,CAAC,YAAY,CACjB,IAAI,GAAE,OAAO,CAAC;QACZ,KAAK,EAAE,MAAM,CAAC;QACd,MAAM,CAAC,EAAE,MAAM,CAAC;QAChB,OAAO,CAAC,EAAE,MAAM,CAAC;QACjB,MAAM,EAAE,MAAM,CAAC;QACf,QAAQ,EAAE,MAAM,CAAC;QACjB,cAAc,EAAE,OAAO,CAAC;KACzB,CAAM,GACN,GAAG;IA6CA,UAAU,CAAC,MAAM,EAAE,WAAW,EAAE,QAAQ,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,WAAW,CAAC;IAgClF,2EAA2E;IAC3E,MAAM,IAAI,GAAG,CAAC,YAAY;IAIpB,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;CAG7B"}
|
package/dist/stt.js
CHANGED
|
@@ -10,6 +10,7 @@ class STT extends stt.STT {
|
|
|
10
10
|
#opts;
|
|
11
11
|
#client;
|
|
12
12
|
label = "openai.STT";
|
|
13
|
+
abortController = new AbortController();
|
|
13
14
|
/**
|
|
14
15
|
* Create a new instance of OpenAI STT.
|
|
15
16
|
*
|
|
@@ -97,13 +98,18 @@ class STT extends stt.STT {
|
|
|
97
98
|
const config = this.#sanitizeOptions(language);
|
|
98
99
|
buffer = mergeFrames(buffer);
|
|
99
100
|
const file = new File([this.#createWav(buffer)], "audio.wav", { type: "audio/wav" });
|
|
100
|
-
const resp = await this.#client.audio.transcriptions.create(
|
|
101
|
-
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
|
|
105
|
-
|
|
106
|
-
|
|
101
|
+
const resp = await this.#client.audio.transcriptions.create(
|
|
102
|
+
{
|
|
103
|
+
file,
|
|
104
|
+
model: this.#opts.model,
|
|
105
|
+
language: config.language,
|
|
106
|
+
prompt: config.prompt,
|
|
107
|
+
response_format: "json"
|
|
108
|
+
},
|
|
109
|
+
{
|
|
110
|
+
signal: this.abortController.signal
|
|
111
|
+
}
|
|
112
|
+
);
|
|
107
113
|
return {
|
|
108
114
|
type: stt.SpeechEventType.FINAL_TRANSCRIPT,
|
|
109
115
|
alternatives: [
|
|
@@ -121,6 +127,9 @@ class STT extends stt.STT {
|
|
|
121
127
|
stream() {
|
|
122
128
|
throw new Error("Streaming is not supported on OpenAI STT");
|
|
123
129
|
}
|
|
130
|
+
async close() {
|
|
131
|
+
this.abortController.abort();
|
|
132
|
+
}
|
|
124
133
|
}
|
|
125
134
|
export {
|
|
126
135
|
STT
|
package/dist/stt.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["../src/stt.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { type AudioBuffer, mergeFrames, stt } from '@livekit/agents';\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport { OpenAI } from 'openai';\nimport type { GroqAudioModels, WhisperModels } from './models.js';\n\nexport interface STTOptions {\n apiKey?: string;\n language: string;\n prompt?: string;\n detectLanguage: boolean;\n model: WhisperModels | string;\n baseURL?: string;\n client?: OpenAI;\n}\n\nconst defaultSTTOptions: STTOptions = {\n apiKey: process.env.OPENAI_API_KEY,\n language: 'en',\n detectLanguage: false,\n model: 'whisper-1',\n};\n\nexport class STT extends stt.STT {\n #opts: STTOptions;\n #client: OpenAI;\n label = 'openai.STT';\n\n /**\n * Create a new instance of OpenAI STT.\n *\n * @remarks\n * `apiKey` must be set to your OpenAI API key, either using the argument or by setting the\n * `OPENAI_API_KEY` environment variable.\n */\n constructor(opts: Partial<STTOptions> = defaultSTTOptions) {\n super({ streaming: false, interimResults: false });\n\n this.#opts = { ...defaultSTTOptions, ...opts };\n if (this.#opts.apiKey === undefined) {\n throw new Error('OpenAI API key is required, whether as an argument or as $OPENAI_API_KEY');\n }\n\n this.#client =\n this.#opts.client ||\n new OpenAI({\n baseURL: opts.baseURL,\n apiKey: opts.apiKey,\n });\n }\n\n /**\n * Create a new instance of Groq STT.\n *\n * @remarks\n * `apiKey` must be set to your Groq API key, either using the argument or by setting the\n * `GROQ_API_KEY` environment variable.\n */\n static withGroq(\n opts: Partial<{\n model: string | GroqAudioModels;\n apiKey?: string;\n baseURL?: string;\n client: OpenAI;\n language: string;\n detectLanguage: boolean;\n }> = {},\n ): STT {\n opts.apiKey = opts.apiKey || process.env.GROQ_API_KEY;\n if (opts.apiKey === undefined) {\n throw new Error('Groq API key is required, whether as an argument or as $GROQ_API_KEY');\n }\n\n return new STT({\n model: 'whisper-large-v3-turbo',\n baseURL: 'https://api.groq.com/openai/v1',\n ...opts,\n });\n }\n\n /**\n * Create a new instance of OVHcloud AI Endpoints STT.\n *\n * @remarks\n * `apiKey` must be set to your OVHcloud AI Endpoints API key, either using the argument or by setting the\n * `OVHCLOUD_API_KEY` environment variable.\n */\n static withOVHcloud(\n opts: Partial<{\n model: string;\n apiKey?: string;\n baseURL?: string;\n client: OpenAI;\n language: string;\n detectLanguage: boolean;\n }> = {},\n ): STT {\n opts.apiKey = opts.apiKey || process.env.OVHCLOUD_API_KEY;\n if (opts.apiKey === undefined) {\n throw new Error(\n 'OVHcloud AI Endpoints API key is required, whether as an argument or as $OVHCLOUD_API_KEY',\n );\n }\n\n return new STT({\n model: 'whisper-large-v3-turbo',\n baseURL: 'https://oai.endpoints.kepler.ai.cloud.ovh.net/v1',\n ...opts,\n });\n }\n\n #sanitizeOptions(language?: string): STTOptions {\n if (language) {\n return { ...this.#opts, language };\n } else {\n return this.#opts;\n }\n }\n\n #createWav(frame: AudioFrame): Buffer {\n const bitsPerSample = 16;\n const byteRate = (frame.sampleRate * frame.channels * bitsPerSample) / 8;\n const blockAlign = (frame.channels * bitsPerSample) / 8;\n\n const header = Buffer.alloc(44);\n header.write('RIFF', 0);\n header.writeUInt32LE(36 + frame.data.byteLength, 4);\n header.write('WAVE', 8);\n header.write('fmt ', 12);\n header.writeUInt32LE(16, 16);\n header.writeUInt16LE(1, 20);\n header.writeUInt16LE(frame.channels, 22);\n header.writeUInt32LE(frame.sampleRate, 24);\n header.writeUInt32LE(byteRate, 28);\n header.writeUInt16LE(blockAlign, 32);\n header.writeUInt16LE(16, 34);\n header.write('data', 36);\n header.writeUInt32LE(frame.data.byteLength, 40);\n return Buffer.concat([header, Buffer.from(frame.data.buffer)]);\n }\n\n async _recognize(buffer: AudioBuffer, language?: string): Promise<stt.SpeechEvent> {\n const config = this.#sanitizeOptions(language);\n buffer = mergeFrames(buffer);\n const file = new File([this.#createWav(buffer)], 'audio.wav', { type: 'audio/wav' });\n const resp = await this.#client.audio.transcriptions.create({\n
|
|
1
|
+
{"version":3,"sources":["../src/stt.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport { type AudioBuffer, mergeFrames, stt } from '@livekit/agents';\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport { OpenAI } from 'openai';\nimport type { GroqAudioModels, WhisperModels } from './models.js';\n\nexport interface STTOptions {\n apiKey?: string;\n language: string;\n prompt?: string;\n detectLanguage: boolean;\n model: WhisperModels | string;\n baseURL?: string;\n client?: OpenAI;\n}\n\nconst defaultSTTOptions: STTOptions = {\n apiKey: process.env.OPENAI_API_KEY,\n language: 'en',\n detectLanguage: false,\n model: 'whisper-1',\n};\n\nexport class STT extends stt.STT {\n #opts: STTOptions;\n #client: OpenAI;\n label = 'openai.STT';\n private abortController = new AbortController();\n\n /**\n * Create a new instance of OpenAI STT.\n *\n * @remarks\n * `apiKey` must be set to your OpenAI API key, either using the argument or by setting the\n * `OPENAI_API_KEY` environment variable.\n */\n constructor(opts: Partial<STTOptions> = defaultSTTOptions) {\n super({ streaming: false, interimResults: false });\n\n this.#opts = { ...defaultSTTOptions, ...opts };\n if (this.#opts.apiKey === undefined) {\n throw new Error('OpenAI API key is required, whether as an argument or as $OPENAI_API_KEY');\n }\n\n this.#client =\n this.#opts.client ||\n new OpenAI({\n baseURL: opts.baseURL,\n apiKey: opts.apiKey,\n });\n }\n\n /**\n * Create a new instance of Groq STT.\n *\n * @remarks\n * `apiKey` must be set to your Groq API key, either using the argument or by setting the\n * `GROQ_API_KEY` environment variable.\n */\n static withGroq(\n opts: Partial<{\n model: string | GroqAudioModels;\n apiKey?: string;\n baseURL?: string;\n client: OpenAI;\n language: string;\n detectLanguage: boolean;\n }> = {},\n ): STT {\n opts.apiKey = opts.apiKey || process.env.GROQ_API_KEY;\n if (opts.apiKey === undefined) {\n throw new Error('Groq API key is required, whether as an argument or as $GROQ_API_KEY');\n }\n\n return new STT({\n model: 'whisper-large-v3-turbo',\n baseURL: 'https://api.groq.com/openai/v1',\n ...opts,\n });\n }\n\n /**\n * Create a new instance of OVHcloud AI Endpoints STT.\n *\n * @remarks\n * `apiKey` must be set to your OVHcloud AI Endpoints API key, either using the argument or by setting the\n * `OVHCLOUD_API_KEY` environment variable.\n */\n static withOVHcloud(\n opts: Partial<{\n model: string;\n apiKey?: string;\n baseURL?: string;\n client: OpenAI;\n language: string;\n detectLanguage: boolean;\n }> = {},\n ): STT {\n opts.apiKey = opts.apiKey || process.env.OVHCLOUD_API_KEY;\n if (opts.apiKey === undefined) {\n throw new Error(\n 'OVHcloud AI Endpoints API key is required, whether as an argument or as $OVHCLOUD_API_KEY',\n );\n }\n\n return new STT({\n model: 'whisper-large-v3-turbo',\n baseURL: 'https://oai.endpoints.kepler.ai.cloud.ovh.net/v1',\n ...opts,\n });\n }\n\n #sanitizeOptions(language?: string): STTOptions {\n if (language) {\n return { ...this.#opts, language };\n } else {\n return this.#opts;\n }\n }\n\n #createWav(frame: AudioFrame): Buffer {\n const bitsPerSample = 16;\n const byteRate = (frame.sampleRate * frame.channels * bitsPerSample) / 8;\n const blockAlign = (frame.channels * bitsPerSample) / 8;\n\n const header = Buffer.alloc(44);\n header.write('RIFF', 0);\n header.writeUInt32LE(36 + frame.data.byteLength, 4);\n header.write('WAVE', 8);\n header.write('fmt ', 12);\n header.writeUInt32LE(16, 16);\n header.writeUInt16LE(1, 20);\n header.writeUInt16LE(frame.channels, 22);\n header.writeUInt32LE(frame.sampleRate, 24);\n header.writeUInt32LE(byteRate, 28);\n header.writeUInt16LE(blockAlign, 32);\n header.writeUInt16LE(16, 34);\n header.write('data', 36);\n header.writeUInt32LE(frame.data.byteLength, 40);\n return Buffer.concat([header, Buffer.from(frame.data.buffer)]);\n }\n\n async _recognize(buffer: AudioBuffer, language?: string): Promise<stt.SpeechEvent> {\n const config = this.#sanitizeOptions(language);\n buffer = mergeFrames(buffer);\n const file = new File([this.#createWav(buffer)], 'audio.wav', { type: 'audio/wav' });\n\n const resp = await this.#client.audio.transcriptions.create(\n {\n file,\n model: this.#opts.model,\n language: config.language,\n prompt: config.prompt,\n response_format: 'json',\n },\n {\n signal: this.abortController.signal,\n },\n );\n\n return {\n type: stt.SpeechEventType.FINAL_TRANSCRIPT,\n alternatives: [\n {\n text: resp.text || '',\n language: language || '',\n startTime: 0,\n endTime: 0,\n confidence: 0,\n },\n ],\n };\n }\n\n /** This method throws an error; streaming is unsupported on OpenAI STT. */\n stream(): stt.SpeechStream {\n throw new Error('Streaming is not supported on OpenAI STT');\n }\n\n async close(): Promise<void> {\n this.abortController.abort();\n }\n}\n"],"mappings":"AAGA,SAA2B,aAAa,WAAW;AAEnD,SAAS,cAAc;AAavB,MAAM,oBAAgC;AAAA,EACpC,QAAQ,QAAQ,IAAI;AAAA,EACpB,UAAU;AAAA,EACV,gBAAgB;AAAA,EAChB,OAAO;AACT;AAEO,MAAM,YAAY,IAAI,IAAI;AAAA,EAC/B;AAAA,EACA;AAAA,EACA,QAAQ;AAAA,EACA,kBAAkB,IAAI,gBAAgB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAS9C,YAAY,OAA4B,mBAAmB;AACzD,UAAM,EAAE,WAAW,OAAO,gBAAgB,MAAM,CAAC;AAEjD,SAAK,QAAQ,EAAE,GAAG,mBAAmB,GAAG,KAAK;AAC7C,QAAI,KAAK,MAAM,WAAW,QAAW;AACnC,YAAM,IAAI,MAAM,0EAA0E;AAAA,IAC5F;AAEA,SAAK,UACH,KAAK,MAAM,UACX,IAAI,OAAO;AAAA,MACT,SAAS,KAAK;AAAA,MACd,QAAQ,KAAK;AAAA,IACf,CAAC;AAAA,EACL;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EASA,OAAO,SACL,OAOK,CAAC,GACD;AACL,SAAK,SAAS,KAAK,UAAU,QAAQ,IAAI;AACzC,QAAI,KAAK,WAAW,QAAW;AAC7B,YAAM,IAAI,MAAM,sEAAsE;AAAA,IACxF;AAEA,WAAO,IAAI,IAAI;AAAA,MACb,OAAO;AAAA,MACP,SAAS;AAAA,MACT,GAAG;AAAA,IACL,CAAC;AAAA,EACH;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EASA,OAAO,aACL,OAOK,CAAC,GACD;AACL,SAAK,SAAS,KAAK,UAAU,QAAQ,IAAI;AACzC,QAAI,KAAK,WAAW,QAAW;AAC7B,YAAM,IAAI;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAEA,WAAO,IAAI,IAAI;AAAA,MACb,OAAO;AAAA,MACP,SAAS;AAAA,MACT,GAAG;AAAA,IACL,CAAC;AAAA,EACH;AAAA,EAEA,iBAAiB,UAA+B;AAC9C,QAAI,UAAU;AACZ,aAAO,EAAE,GAAG,KAAK,OAAO,SAAS;AAAA,IACnC,OAAO;AACL,aAAO,KAAK;AAAA,IACd;AAAA,EACF;AAAA,EAEA,WAAW,OAA2B;AACpC,UAAM,gBAAgB;AACtB,UAAM,WAAY,MAAM,aAAa,MAAM,WAAW,gBAAiB;AACvE,UAAM,aAAc,MAAM,WAAW,gBAAiB;AAEtD,UAAM,SAAS,OAAO,MAAM,EAAE;AAC9B,WAAO,MAAM,QAAQ,CAAC;AACtB,WAAO,cAAc,KAAK,MAAM,KAAK,YAAY,CAAC;AAClD,WAAO,MAAM,QAAQ,CAAC;AACtB,WAAO,MAAM,QAAQ,EAAE;AACvB,WAAO,cAAc,IAAI,EAAE;AAC3B,WAAO,cAAc,GAAG,EAAE;AAC1B,WAAO,cAAc,MAAM,UAAU,EAAE;AACvC,WAAO,cAAc,MAAM,YAAY,EAAE;AACzC,WAAO,cAAc,UAAU,EAAE;AACjC,WAAO,cAAc,YAAY,EAAE;AACnC,WAAO,cAAc,IAAI,EAAE;AAC3B,WAAO,MAAM,QAAQ,EAAE;AACvB,WAAO,cAAc,MAAM,KAAK,YAAY,EAAE;AAC9C,WAAO,OAAO,OAAO,CAAC,QAAQ,OAAO,KAAK,MAAM,KAAK,MAAM,CAAC,CAAC;AAAA,EAC/D;AAAA,EAEA,MAAM,WAAW,QAAqB,UAA6C;AACjF,UAAM,SAAS,KAAK,iBAAiB,QAAQ;AAC7C,aAAS,YAAY,MAAM;AAC3B,UAAM,OAAO,IAAI,KAAK,CAAC,KAAK,WAAW,MAAM,CAAC,GAAG,aAAa,EAAE,MAAM,YAAY,CAAC;AAEnF,UAAM,OAAO,MAAM,KAAK,QAAQ,MAAM,eAAe;AAAA,MACnD;AAAA,QACE;AAAA,QACA,OAAO,KAAK,MAAM;AAAA,QAClB,UAAU,OAAO;AAAA,QACjB,QAAQ,OAAO;AAAA,QACf,iBAAiB;AAAA,MACnB;AAAA,MACA;AAAA,QACE,QAAQ,KAAK,gBAAgB;AAAA,MAC/B;AAAA,IACF;AAEA,WAAO;AAAA,MACL,MAAM,IAAI,gBAAgB;AAAA,MAC1B,cAAc;AAAA,QACZ;AAAA,UACE,MAAM,KAAK,QAAQ;AAAA,UACnB,UAAU,YAAY;AAAA,UACtB,WAAW;AAAA,UACX,SAAS;AAAA,UACT,YAAY;AAAA,QACd;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAAA;AAAA,EAGA,SAA2B;AACzB,UAAM,IAAI,MAAM,0CAA0C;AAAA,EAC5D;AAAA,EAEA,MAAM,QAAuB;AAC3B,SAAK,gBAAgB,MAAM;AAAA,EAC7B;AACF;","names":[]}
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@livekit/agents-plugin-openai",
|
|
3
|
-
"version": "1.0.
|
|
3
|
+
"version": "1.0.23",
|
|
4
4
|
"description": "OpenAI plugin for LiveKit Node Agents",
|
|
5
5
|
"main": "dist/index.js",
|
|
6
6
|
"require": "dist/index.cjs",
|
|
@@ -30,9 +30,9 @@
|
|
|
30
30
|
"@types/ws": "^8.5.10",
|
|
31
31
|
"tsup": "^8.3.5",
|
|
32
32
|
"typescript": "^5.0.0",
|
|
33
|
-
"@livekit/agents": "1.0.
|
|
34
|
-
"@livekit/agents-plugin-silero": "1.0.
|
|
35
|
-
"@livekit/agents-plugins-test": "1.0.
|
|
33
|
+
"@livekit/agents": "1.0.23",
|
|
34
|
+
"@livekit/agents-plugin-silero": "1.0.23",
|
|
35
|
+
"@livekit/agents-plugins-test": "1.0.23"
|
|
36
36
|
},
|
|
37
37
|
"dependencies": {
|
|
38
38
|
"@livekit/mutex": "^1.1.1",
|
|
@@ -41,7 +41,7 @@
|
|
|
41
41
|
},
|
|
42
42
|
"peerDependencies": {
|
|
43
43
|
"@livekit/rtc-node": "^0.13.12",
|
|
44
|
-
"@livekit/agents": "1.0.
|
|
44
|
+
"@livekit/agents": "1.0.23"
|
|
45
45
|
},
|
|
46
46
|
"scripts": {
|
|
47
47
|
"build": "tsup --onSuccess \"pnpm build:types\"",
|
package/src/stt.ts
CHANGED
|
@@ -27,6 +27,7 @@ export class STT extends stt.STT {
|
|
|
27
27
|
#opts: STTOptions;
|
|
28
28
|
#client: OpenAI;
|
|
29
29
|
label = 'openai.STT';
|
|
30
|
+
private abortController = new AbortController();
|
|
30
31
|
|
|
31
32
|
/**
|
|
32
33
|
* Create a new instance of OpenAI STT.
|
|
@@ -145,13 +146,19 @@ export class STT extends stt.STT {
|
|
|
145
146
|
const config = this.#sanitizeOptions(language);
|
|
146
147
|
buffer = mergeFrames(buffer);
|
|
147
148
|
const file = new File([this.#createWav(buffer)], 'audio.wav', { type: 'audio/wav' });
|
|
148
|
-
|
|
149
|
-
|
|
150
|
-
|
|
151
|
-
|
|
152
|
-
|
|
153
|
-
|
|
154
|
-
|
|
149
|
+
|
|
150
|
+
const resp = await this.#client.audio.transcriptions.create(
|
|
151
|
+
{
|
|
152
|
+
file,
|
|
153
|
+
model: this.#opts.model,
|
|
154
|
+
language: config.language,
|
|
155
|
+
prompt: config.prompt,
|
|
156
|
+
response_format: 'json',
|
|
157
|
+
},
|
|
158
|
+
{
|
|
159
|
+
signal: this.abortController.signal,
|
|
160
|
+
},
|
|
161
|
+
);
|
|
155
162
|
|
|
156
163
|
return {
|
|
157
164
|
type: stt.SpeechEventType.FINAL_TRANSCRIPT,
|
|
@@ -171,4 +178,8 @@ export class STT extends stt.STT {
|
|
|
171
178
|
stream(): stt.SpeechStream {
|
|
172
179
|
throw new Error('Streaming is not supported on OpenAI STT');
|
|
173
180
|
}
|
|
181
|
+
|
|
182
|
+
async close(): Promise<void> {
|
|
183
|
+
this.abortController.abort();
|
|
184
|
+
}
|
|
174
185
|
}
|