Update app.py

app.py

@@ -1,6 +1,9 @@
 import gradio as gr
 from TTS.api import TTS
 import numpy as np
+from scipy.io import wavfile
+import tempfile
+import os
 
 # Load the YourTTS model once at startup
 tts = TTS(model_name="tts_models/multilingual/multi-dataset/your_tts", progress_bar=False)
@@ -15,13 +18,21 @@ def generate_speech(reference_audio, text):
     text (str): Text to convert to speech.
     
     Returns:
-    tuple: (audio waveform as numpy array, sample rate)
+    str: Path to the generated audio file
     """
     # Generate speech using the reference audio and text
     wav = tts.tts(text=text, speaker_wav=reference_audio, language="en")
-    # Convert list to numpy array for Gradio
-    wav_np = np.array(wav)
-    return (wav_np, sample_rate)
+    # Convert list to numpy array
+    wav_np = np.array(wav, dtype=np.float32)
+    
+    # Create a temporary file to save the audio
+    temp_file = tempfile.NamedTemporaryFile(suffix=".wav", delete=False)
+    temp_file_path = temp_file.name
+    # Save the audio to the temporary file
+    wavfile.write(temp_file_path, sample_rate, wav_np)
+    temp_file.close()
+    
+    return temp_file_path
 
 # Build the Gradio interface
 with gr.Blocks(title="Voice Cloning TTS") as app: