From 65c83317aa5c692e4177b30429521c467ca12ab1 Mon Sep 17 00:00:00 2001
From: Danny Avila <danny@librechat.ai>
Date: Fri, 5 Sep 2025 12:01:00 -0400
Subject: [PATCH] =?UTF-8?q?=F0=9F=97=A3=EF=B8=8F=20feat:=20Language=20Supp?=
 =?UTF-8?q?ort=20for=20OpenAI=20Speech-to-Text=20(#9470)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 api/server/services/Files/Audio/STTService.js | 33 ++++++++++++++++---
 .../hooks/Input/useSpeechToTextExternal.ts    |  4 +++
 2 files changed, 32 insertions(+), 5 deletions(-)

diff --git a/api/server/services/Files/Audio/STTService.js b/api/server/services/Files/Audio/STTService.js
index 95eea62ab..60d6a48a1 100644
--- a/api/server/services/Files/Audio/STTService.js
+++ b/api/server/services/Files/Audio/STTService.js
@@ -159,9 +159,11 @@ class STTService {
    * Prepares the request for the OpenAI STT provider.
    * @param {Object} sttSchema - The STT schema for OpenAI.
    * @param {Stream} audioReadStream - The audio data to be transcribed.
+   * @param {Object} audioFile - The audio file object (unused in OpenAI provider).
+   * @param {string} language - The language code for the transcription.
    * @returns {Array} An array containing the URL, data, and headers for the request.
    */
-  openAIProvider(sttSchema, audioReadStream) {
+  openAIProvider(sttSchema, audioReadStream, audioFile, language) {
     const url = sttSchema?.url || 'https://api.openai.com/v1/audio/transcriptions';
     const apiKey = extractEnvVariable(sttSchema.apiKey) || '';
 
@@ -170,6 +172,12 @@ class STTService {
       model: sttSchema.model,
     };
 
+    if (language) {
+      /** Converted locale code (e.g., "en-US") to ISO-639-1 format (e.g., "en") */
+      const isoLanguage = language.split('-')[0];
+      data.language = isoLanguage;
+    }
+
     const headers = {
       'Content-Type': 'multipart/form-data',
       ...(apiKey && { Authorization: `Bearer ${apiKey}` }),
@@ -184,10 +192,11 @@ class STTService {
    * @param {Object} sttSchema - The STT schema for Azure OpenAI.
    * @param {Buffer} audioBuffer - The audio data to be transcribed.
    * @param {Object} audioFile - The audio file object containing originalname, mimetype, and size.
+   * @param {string} language - The language code for the transcription.
    * @returns {Array} An array containing the URL, data, and headers for the request.
    * @throws {Error} If the audio file size exceeds 25MB or the audio file format is not accepted.
    */
-  azureOpenAIProvider(sttSchema, audioBuffer, audioFile) {
+  azureOpenAIProvider(sttSchema, audioBuffer, audioFile, language) {
     const url = `${genAzureEndpoint({
       azureOpenAIApiInstanceName: extractEnvVariable(sttSchema?.instanceName),
       azureOpenAIApiDeploymentName: extractEnvVariable(sttSchema?.deploymentName),
@@ -211,6 +220,12 @@ class STTService {
       contentType: audioFile.mimetype,
     });
 
+    if (language) {
+      /** Converted locale code (e.g., "en-US") to ISO-639-1 format (e.g., "en") */
+      const isoLanguage = language.split('-')[0];
+      formData.append('language', isoLanguage);
+    }
+
     const headers = {
       'Content-Type': 'multipart/form-data',
       ...(apiKey && { 'api-key': apiKey }),
@@ -229,10 +244,11 @@ class STTService {
    * @param {Object} requestData - The data required for the STT request.
    * @param {Buffer} requestData.audioBuffer - The audio data to be transcribed.
    * @param {Object} requestData.audioFile - The audio file object containing originalname, mimetype, and size.
+   * @param {string} requestData.language - The language code for the transcription.
    * @returns {Promise<string>} A promise that resolves to the transcribed text.
    * @throws {Error} If the provider is invalid, the response status is not 200, or the response data is missing.
    */
-  async sttRequest(provider, sttSchema, { audioBuffer, audioFile }) {
+  async sttRequest(provider, sttSchema, { audioBuffer, audioFile, language }) {
     const strategy = this.providerStrategies[provider];
     if (!strategy) {
       throw new Error('Invalid provider');
@@ -243,7 +259,13 @@ class STTService {
     const audioReadStream = Readable.from(audioBuffer);
     audioReadStream.path = `audio.${fileExtension}`;
 
-    const [url, data, headers] = strategy.call(this, sttSchema, audioReadStream, audioFile);
+    const [url, data, headers] = strategy.call(
+      this,
+      sttSchema,
+      audioReadStream,
+      audioFile,
+      language,
+    );
 
     try {
       const response = await axios.post(url, data, { headers });
@@ -284,7 +306,8 @@ class STTService {
 
     try {
       const [provider, sttSchema] = await this.getProviderSchema(req);
-      const text = await this.sttRequest(provider, sttSchema, { audioBuffer, audioFile });
+      const language = req.body?.language || '';
+      const text = await this.sttRequest(provider, sttSchema, { audioBuffer, audioFile, language });
       res.json({ text });
     } catch (error) {
       logger.error('An error occurred while processing the audio:', error);
diff --git a/client/src/hooks/Input/useSpeechToTextExternal.ts b/client/src/hooks/Input/useSpeechToTextExternal.ts
index 6571a0911..0376279ce 100644
--- a/client/src/hooks/Input/useSpeechToTextExternal.ts
+++ b/client/src/hooks/Input/useSpeechToTextExternal.ts
@@ -25,6 +25,7 @@ const useSpeechToTextExternal = (
 
   const [minDecibels] = useRecoilState(store.decibelValue);
   const [autoSendText] = useRecoilState(store.autoSendText);
+  const [languageSTT] = useRecoilState<string>(store.languageSTT);
   const [speechToText] = useRecoilState<boolean>(store.speechToText);
   const [autoTranscribeAudio] = useRecoilState<boolean>(store.autoTranscribeAudio);
 
@@ -121,6 +122,9 @@ const useSpeechToTextExternal = (
 
       const formData = new FormData();
       formData.append('audio', audioBlob, `audio.${fileExtension}`);
+      if (languageSTT) {
+        formData.append('language', languageSTT);
+      }
       setIsRequestBeingMade(true);
       cleanup();
       processAudio(formData);