Update app.py

app.py

@@ -54,21 +54,26 @@ HTML = """
       } else if (data.audio) {
         status.textContent = 'Audio generated successfully!';
         
-        // Create audio element
+        // Create audio element with correct MIME type
         const audio = document.createElement('audio');
         audio.controls = true;
         audio.style.width = '100%';
-        audio.src = 'data:audio/mp3;base64,' + data.audio;
+        const mimeType = data.mime_type || 'audio/wav';
+        audio.src = `data:${mimeType};base64,${data.audio}`;
         audioContainer.appendChild(audio);
         
         // Create download link
         const downloadLink = document.createElement('a');
         downloadLink.href = audio.src;
-        downloadLink.download = 'gemini_audio.mp3';
+        const ext = mimeType.includes('mp3') ? 'mp3' : 'wav';
+        downloadLink.download = `gemini_audio.${ext}`;
         downloadLink.textContent = 'Download Audio';
         downloadLink.style.display = 'block';
         downloadLink.style.marginTop = '1rem';
         audioContainer.appendChild(downloadLink);
+        
+        // Auto-play
+        audio.play().catch(err => console.log('Autoplay prevented:', err));
       } else {
         status.textContent = 'Unexpected response payload';
       }
@@ -82,12 +87,13 @@ HTML = """
 </html>
 """
 
-def generate_audio_from_gemini(prompt: str) -> bytes:
+def generate_audio_from_gemini(prompt: str) -> tuple[bytes, str]:
     client = genai.Client(
         api_key="AIzaSyDolbPUZBPUPvQUu-RGktJmvnUpkcEKIYo",
     )
     
-    model = "gemini-2.5-flash-preview-tts" #"gemini-2.0-flash"
+    model = "gemini-2.0-flash-exp"
+    # model = "gemini-2.5-flash-preview-tts"
     contents = [
         types.Content(
             role="user",
@@ -101,7 +107,7 @@ def generate_audio_from_gemini(prompt: str) -> bytes:
         speech_config=types.SpeechConfig(
             voice_config=types.VoiceConfig(
                 prebuilt_voice_config=types.PrebuiltVoiceConfig(
-                    voice_name="Puck"  # Options: Puck, Charon, Kore, Fenrir, Aoede
+                    voice_name="Puck"
                 )
             )
         )
@@ -114,11 +120,18 @@ def generate_audio_from_gemini(prompt: str) -> bytes:
         config=config,
     )
     
+    # Log response structure for debugging
+    app.logger.info(f"Response candidates: {len(response.candidates)}")
+    if response.candidates:
+        app.logger.info(f"Parts in response: {len(response.candidates[0].content.parts)}")
+    
     # Extract audio data from response
-    # The audio is in response.candidates[0].content.parts
     for part in response.candidates[0].content.parts:
-        if part.inline_data and part.inline_data.mime_type.startswith("audio/"):
-            return part.inline_data.data
+        if hasattr(part, 'inline_data') and part.inline_data:
+            mime_type = part.inline_data.mime_type
+            audio_data = part.inline_data.data
+            app.logger.info(f"Found audio: mime_type={mime_type}, size={len(audio_data)} bytes")
+            return audio_data, mime_type
     
     raise ValueError("No audio data found in response")
     
@@ -134,11 +147,11 @@ def gen():
     if not prompt:
         return jsonify({"error": "No prompt provided"}), 400
     try:
-        audio_bytes = generate_audio_from_gemini(prompt)
+        audio_bytes, mime_type = generate_audio_from_gemini(prompt)
         # Convert audio bytes to base64 for transmission
         audio_base64 = base64.b64encode(audio_bytes).decode('utf-8')
-        app.logger.info("Audio generation succeeded, size=%d bytes", len(audio_bytes))
-        return jsonify({"audio": audio_base64})
+        app.logger.info("Audio generation succeeded, size=%d bytes, mime=%s", len(audio_bytes), mime_type)
+        return jsonify({"audio": audio_base64, "mime_type": mime_type})
     except Exception as e:
         app.logger.exception("Audio generation failed")
         return jsonify({"error": str(e)}), 500
@@ -146,4 +159,4 @@ def gen():
         
 if __name__ == "__main__":
     port = int(os.environ.get("PORT", 7860))
-    app.run(host="0.0.0.0", port=port)
+    app.run(host="0.0.0.0", port=port, debug=True)