update-tansc

src/server/main.py

@@ -816,39 +816,54 @@ async def speech_to_speech(
     try:
         encrypted_content = await file.read()
         file_content = decrypt_data(encrypted_content, session_key)
-
-
-
         with tempfile.NamedTemporaryFile(delete=False, suffix=file.filename) as temp_file:
-        # Write the decrypted content to the temp file
-            #decrypted_content = await file.read()  # Assuming decrypted_content is the file content
             temp_file.write(file_content)
             temp_file_path = temp_file.name
 
+        try:
+            # Transcribe audio
+            response = dwani.ASR.transcribe(file_path=temp_file_path, language=decrypted_language)
+            transcription = response.get("text", "")
+            if not transcription:
+                logger.error("Transcription is empty")
+                raise HTTPException(status_code=400, detail="Transcription failed or returned empty text")
 
-        response = dwani.ASR.transcribe(file_path=temp_file_path, language=decrypted_language)
-
-        transcription = response.get("text","")
-
-        chat_response = dwani.Chat.create(prompt=transcription, src_lang=decrypted_language, tgt_lang=decrypted_language)
-        
-
-        response_text = chat_response.get("response", "")
-
-        response = dwani.Audio.speech(input=response_text, response_format="mp3")
-
-        headers = {
-            "Content-Disposition": f"inline; filename=\"speech.mp3\"",
-            "Cache-Control": "no-cache",
-            "Content-Type": "audio/mp3"
-        }
+            # Debug inputs
+            logger.debug(f"Transcription: {transcription}, Language: {decrypted_language}")
 
-        return StreamingResponse(
-        bytes_iterator(response),
-            media_type="audio/mp3",
-            headers=headers
-        )
+            # Chat processing
+            try:
+                chat_response = dwani.Chat.create(
+                    prompt=transcription,
+                    src_lang=decrypted_language,
+                    tgt_lang=decrypted_language
+                )
+            except dwani.exceptions.DhwaniAPIError as e:
+                logger.error(f"Chat API failed: {str(e)}")
+                raise HTTPException(status_code=502, detail=f"Chat service error: {str(e)}")
+
+            response_text = chat_response.get("response", "")
+            if not response_text:
+                logger.error("Chat response is empty")
+                raise HTTPException(status_code=500, detail="Chat service returned empty response")
+
+            # Generate audio
+            response = dwani.Audio.speech(input=response_text, response_format="mp3")
+
+            headers = {
+                "Content-Disposition": f"inline; filename=\"speech.mp3\"",
+                "Cache-Control": "no-cache",
+                "Content-Type": "audio/mp3"
+            }
 
+            return StreamingResponse(
+                bytes_iterator(response),
+                media_type="audio/mp3",
+                headers=headers
+            )
+        finally:
+            # Clean up temporary file
+            os.unlink(temp_file_path)
     except requests.Timeout:
         logger.error("External speech-to-speech API timed out", extra={"user_id": user_id})
         raise HTTPException(status_code=504, detail="External API timeout")