211 lines
4.9 KiB
TypeScript
211 lines
4.9 KiB
TypeScript
import { config } from '../../utils/config';
|
|
import { TTSModule, TTSResponse } from '../tts';
|
|
|
|
import * as https from 'https';
|
|
|
|
const ELEVENLABS_API_ENDPOINT = 'api.elevenlabs.io';
|
|
|
|
interface ElevenLabsVoice {
|
|
voice_id: string;
|
|
name: string;
|
|
// ...
|
|
}
|
|
|
|
interface ElevenLabsModel {
|
|
model_id: string;
|
|
name: string;
|
|
// ...
|
|
}
|
|
|
|
interface ElevenLabsVoicesRes {
|
|
voices?: Array<ElevenLabsVoice>;
|
|
}
|
|
|
|
interface ElevenLabsVoiceSettings {
|
|
stability: number;
|
|
similarity_boost: number;
|
|
style: number;
|
|
speed: number;
|
|
user_speaker_boost: boolean;
|
|
}
|
|
|
|
interface ElevenLabsStreamRequest {
|
|
text: string;
|
|
model_id: string;
|
|
voice_settings: ElevenLabsVoiceSettings;
|
|
}
|
|
|
|
export class ElevenLabsTTS implements TTSModule {
|
|
private voices: Array<ElevenLabsVoice> | undefined = undefined;
|
|
private models: Array<ElevenLabsModel> | undefined = undefined;
|
|
|
|
public name: string = 'ElevenLabs';
|
|
|
|
public settings: ElevenLabsVoiceSettings;
|
|
public modelId: string;
|
|
|
|
public static readonly DEFAULT_SETTINGS: ElevenLabsVoiceSettings = {
|
|
stability: 0.5,
|
|
similarity_boost: 0.5,
|
|
style: 0.0,
|
|
speed: 1.0,
|
|
user_speaker_boost: true
|
|
};
|
|
|
|
constructor() {
|
|
if (this.canBeUsed()) {
|
|
this.fetchVoices();
|
|
this.fetchModels();
|
|
}
|
|
|
|
this.settings = ElevenLabsTTS.DEFAULT_SETTINGS;
|
|
|
|
this.modelId = 'eleven_flash_v2_5';
|
|
|
|
this.setSettings = this.setSettings.bind(this);
|
|
this.setModel = this.setModel.bind(this);
|
|
this.getModels = this.getModels.bind(this);
|
|
}
|
|
|
|
/*
|
|
TTSModule methods
|
|
*/
|
|
async getVoices(): Promise<Array<string> | undefined> {
|
|
if (this.voices) return this.voices.map((voice) => voice.name);
|
|
}
|
|
|
|
async generate(voice: string, text: string): Promise<TTSResponse> {
|
|
if (!this.voices) return {};
|
|
|
|
const voiceData = this.voices.find((entry) => entry.name === voice);
|
|
if (!voiceData) return { error: 'Invalid voice' };
|
|
|
|
const options: https.RequestOptions = {
|
|
hostname: ELEVENLABS_API_ENDPOINT,
|
|
path: `/v1/text-to-speech/${voiceData.voice_id}/stream`,
|
|
method: 'POST',
|
|
headers: {
|
|
Accept: 'application/json',
|
|
'xi-api-key': config.tts_elevenlabs_key,
|
|
'Content-Type': 'application/json'
|
|
}
|
|
};
|
|
|
|
const body: ElevenLabsStreamRequest = {
|
|
text: text,
|
|
model_id: this.modelId,
|
|
voice_settings: this.settings
|
|
};
|
|
|
|
return new Promise((resolve) => {
|
|
const req = https.request(options, (res) => {
|
|
const chunks: Buffer[] = [];
|
|
res.on('data', (chunk) => chunks.push(chunk));
|
|
res.on('end', () => {
|
|
resolve({
|
|
data: Buffer.concat(chunks)
|
|
});
|
|
});
|
|
});
|
|
|
|
req.on('error', (error) => resolve({ error: error.message }));
|
|
|
|
req.write(JSON.stringify(body));
|
|
req.end();
|
|
});
|
|
}
|
|
|
|
canBeUsed(): boolean {
|
|
return config.tts_elevenlabs_key != undefined;
|
|
}
|
|
|
|
/*
|
|
ElevenLabs specific methods
|
|
*/
|
|
public setSettings(settings: Partial<ElevenLabsVoiceSettings>) {
|
|
this.settings = { ...this.settings, ...settings };
|
|
}
|
|
|
|
public setModel(name: string) {
|
|
if (!this.models) return;
|
|
|
|
const model = this.models.find((mod) => mod.name == name);
|
|
if (!model) return;
|
|
|
|
this.modelId = model.model_id;
|
|
}
|
|
|
|
public getModels(): Array<string> {
|
|
if (!this.models) return [];
|
|
return this.models.map((mod) => mod.model_id);
|
|
}
|
|
|
|
private async fetchVoices(): Promise<void> {
|
|
const opt: https.RequestOptions = {
|
|
hostname: ELEVENLABS_API_ENDPOINT,
|
|
path: '/v2/voices',
|
|
method: 'GET',
|
|
headers: {
|
|
Accept: 'application/json',
|
|
'xi-api-key': config.tts_elevenlabs_key,
|
|
'Content-Type': 'application/json'
|
|
}
|
|
};
|
|
|
|
return new Promise((resolve) => {
|
|
const req = https.get(opt, (res) => {
|
|
const chunks: Buffer[] = [];
|
|
res.on('data', (chunk) => chunks.push(chunk));
|
|
res.on('end', () => {
|
|
const voicesJSON = Buffer.concat(chunks).toString('utf-8');
|
|
const voicesParsed = JSON.parse(voicesJSON) as ElevenLabsVoicesRes;
|
|
if (!voicesParsed.voices) {
|
|
console.error('ElevenLabs voice fetch responded:', voicesJSON);
|
|
throw new Error('Failed to get ElevenLabs voices');
|
|
}
|
|
this.voices = voicesParsed.voices;
|
|
resolve();
|
|
});
|
|
});
|
|
|
|
req.on('error', (err) => {
|
|
console.error('Failed to get ElevenLabs voices:', err);
|
|
throw err;
|
|
});
|
|
});
|
|
}
|
|
|
|
private async fetchModels(): Promise<void> {
|
|
const opt: https.RequestOptions = {
|
|
hostname: ELEVENLABS_API_ENDPOINT,
|
|
path: '/v1/models',
|
|
method: 'GET',
|
|
headers: {
|
|
Accept: 'application/json',
|
|
'xi-api-key': config.tts_elevenlabs_key,
|
|
'Content-Type': 'application/json'
|
|
}
|
|
};
|
|
|
|
return new Promise((resolve) => {
|
|
const req = https.get(opt, (res) => {
|
|
const chunks: Buffer[] = [];
|
|
res.on('data', (chunk) => chunks.push(chunk));
|
|
res.on('end', () => {
|
|
const modelsJSON = Buffer.concat(chunks).toString('utf-8');
|
|
const modelsParsed = JSON.parse(modelsJSON) as Array<ElevenLabsModel>;
|
|
this.models = modelsParsed;
|
|
resolve();
|
|
});
|
|
});
|
|
|
|
req.on('error', (err) => {
|
|
console.error('Failed to get ElevenLabs models:', err);
|
|
throw err;
|
|
});
|
|
});
|
|
}
|
|
}
|
|
|
|
export default new ElevenLabsTTS();
|