Automatic language detection for TTS

server/pyproject.toml

@@ -12,6 +12,7 @@ grpcio = "^1.64.0"
 protobuf = "^5.27.0"
 pyttsx3 = "^2.90"
 pydub = "^0.25.1"
+langdetect = "^1.0.9"
 
 
 [build-system]

server/src/server.py

@@ -1,25 +1,44 @@
-import grpc
-from concurrent import futures
+import io
 import time
+from concurrent import futures
 import pyttsx3
+import grpc
 from pydub import AudioSegment
-import io
+from langdetect import detect
 import server_pb2
 import server_pb2_grpc
 
+
 class TTSService(server_pb2_grpc.TTSServicer):
-    def Synthesize(self, request, context):
-        engine = pyttsx3.init()
+    def __init__(self):
+        self.engine = pyttsx3.init()
+
+    def detect_language(self, text):
+        try:
+            lang = detect(text)
+            return lang
+        except Exception as e:
+            print(f"Error detecting language: {e}")
+            return "en"
+
+    def select_voice(self, engine, lang):
         voices = engine.getProperty('voices')
-        engine.setProperty('voice', voices[1].id)
-        engine.setProperty('rate', 150)
+        for voice in voices:
+            if lang in voice.id.lower():
+                return voice.id
+        return voices[0].id
+
+    def Synthesize(self, request, context):
+        detected_lang = self.detect_language(request.text)
+        voice_id = self.select_voice(self.engine, detected_lang)
+
+        self.engine.setProperty('voice', voice_id)
+        self.engine.setProperty('rate', 150)
 
-        # Save TTS output to a temporary WAV file
         temp_wav_file = 'temp.wav'
-        engine.save_to_file(request.text, temp_wav_file)
-        engine.runAndWait()
+        self.engine.save_to_file(request.text, temp_wav_file)
+        self.engine.runAndWait()
 
-        # Convert WAV file to MP3 and load into byte stream
         audio = AudioSegment.from_wav(temp_wav_file)
         buffer = io.BytesIO()
         audio.export(buffer, format='mp3')
@@ -28,6 +47,7 @@ def Synthesize(self, request, context):
 
         return server_pb2.TTSResponse(track=audio_bytes)
 
+
 def serve():
     server = grpc.server(futures.ThreadPoolExecutor(max_workers=10))
     server_pb2_grpc.add_TTSServicer_to_server(TTSService(), server)
@@ -41,5 +61,6 @@ def serve():
         server.stop(0)
         print("Server stopped.")
 
+
 if __name__ == '__main__':
     serve()