Update app.py

app.py

@@ -37,21 +37,26 @@ def text_to_speech_kokoro(text: str, voice: str = 'af_heart', speed: float = 1.0
         speed: The speed of the speech (default is 1.0).
     
     Returns:
-        The filename of the generated audio file.
+        An AgentAudio object with the relative URL to the generated audio file.
     """
     try:
         # Generate speech audio
         generator = pipeline(text, voice=voice, speed=speed, split_pattern=r'\n+')
-        audio_files = []
-
-        # Save each audio segment to a file
-        for i, (gs, ps, audio) in enumerate(generator):
-            os.makedirs("static", exist_ok=True)
-            filename = os.path.join("static", f'output_{i}.wav')
-            sf.write(filename, audio, 24000)
-            audio_files.append(filename)
-
-        return f"Generated {len(audio_files)} audio file(s): {', '.join(audio_files)}"
+        audio_segments = []
+        for _, _, audio in generator:
+            audio_segments.append(audio)
+        if not audio_segments:
+            raise ValueError("No audio generated.")
+        # Concatenate segments into one audio array
+        full_audio = np.concatenate(audio_segments)
+        sample_rate = 24000  # Kokoro outputs at 24 kHz
+        # Ensure the static folder exists and save the file there
+        os.makedirs("static", exist_ok=True)
+        filename = os.path.join("static", "output.wav")
+        sf.write(filename, full_audio, sample_rate)
+        # Return an AgentAudio object pointing to the relative URL of the audio file
+        from smolagents.agent_types import AgentAudio
+        return AgentAudio(f"/static/output.wav")
     except Exception as e:
         return f"Error generating speech: {str(e)}"