Andreas/LibreChat
1
0
Fork 0
mirror of https://github.com/danny-avila/LibreChat.git synced 2026-01-29 05:36:13 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 9

🌊 feat: update Deepgram SDK integration for STT and remove unused TTS provider

Browse source
This commit is contained in:
Marco Beretta 2024-11-23 12:20:17 +01:00
parent 25d51eff31
commit 5eabd2493c
No known key found for this signature in database
GPG key ID: D918033D8E74CC11
2 changed files with 10 additions and 48 deletions
Download patch file Download diff file Expand all files Collapse all files

12
api/server/services/Files/Audio/STTService.js
View file

@ -158,7 +158,15 @@ class STTService {
return [url, formData, { ...headers, ...formData.getHeaders() }];
}
async deepgramSDKProvider(sttSchema, audioReadStream, audioFile) {
/**
* Transcribes audio using the Deepgram SDK.
* @async
* @param {Object} sttSchema - The STT schema for Deepgram.
* @param {Stream} audioReadStream - The audio data to be transcribed.
* @returns {Promise<string>} A promise that resolves to the transcribed text.
* @throws {Error} If the transcription fails.
*/
async deepgramSDKProvider(sttSchema, audioReadStream) {
const apiKey = extractEnvVariable(sttSchema.apiKey) || '';
const deepgram = createClient(apiKey);
@ -194,7 +202,7 @@ class STTService {
[configOptions].forEach(this.removeUndefined);
const { result, error } = await deepgram.listen.prerecorded.transcribeFile(
Buffer.isBuffer(audioFile) ? audioFile : audioReadStream,
audioReadStream,
configOptions,
);

46
api/server/services/Files/Audio/TTSService.js
View file

@ -248,52 +248,6 @@ class TTSService {
return [url, data, headers];
}
deepgramProvider(ttsSchema, input, voice) {
const baseUrl = ttsSchema?.url || 'https://api.deepgram.com/v1/speak';
const params = {
model: ttsSchema.model,
voice: voice,
language: ttsSchema.language,
};
const queryParams = Object.entries(params)
.filter(([, value]) => value)
.map(([key, value]) => `${key}=${value}`)
.join('&');
const url = queryParams ? `${baseUrl}?${queryParams}` : baseUrl;
if (
ttsSchema?.voices &&
ttsSchema.voices.length > 0 &&
!ttsSchema.voices.includes(voice) &&
!ttsSchema.voices.includes('ALL')
) {
throw new Error(`Voice ${voice} is not available.`);
}
const data = {
input,
model: ttsSchema?.voices && ttsSchema.voices.length > 0 ? voice : undefined,
language: ttsSchema?.language,
media_settings: {
bit_rate: ttsSchema?.media_settings?.bit_rate,
sample_rate: ttsSchema?.media_settings?.sample_rate,
},
};
const headers = {
'Content-Type': 'application/json',
Authorization: `Bearer ${extractEnvVariable(ttsSchema?.apiKey)}`,
};
if (extractEnvVariable(ttsSchema.apiKey) === '') {
delete headers.Authorization;
}
return [url, data, headers];
}
/**
* Sends a TTS request to the specified provider.
* @async