Update app.py

app.py

@@ -21,21 +21,35 @@ vocoder = vocoder.to(device)
 speaker_embedding = torch.zeros(1, 512).to(device)
 
 def tts_generate(text):
-    # Preprocess input
-    inputs = processor(text=text, return_tensors="pt").to(device)
+    print(f"📝 Input text: {text}")
+    try:
+        # Preprocess input
+        print("🔄 Processing input...")
+        inputs = processor(text=text, return_tensors="pt").to(device)
+        print("✅ Text processed.")
 
-    # Generate mel spectrogram
-    with torch.no_grad():
-        mel = model.generate_speech(inputs["input_ids"], speaker_embedding)
+        # Generate mel spectrogram
+        print("🎤 Generating speech...")
+        with torch.no_grad():
+            mel = model.generate_speech(inputs["input_ids"], speaker_embedding)
+        print("✅ Mel spectrogram generated.")
 
-    # Convert mel spectrogram to waveform
-    waveform = vocoder(mel)
-    waveform = waveform.cpu()
+        # Convert mel spectrogram to waveform
+        print("🎚️ Vocoding waveform...")
+        waveform = vocoder(mel)
+        waveform = waveform.cpu()
+        print("✅ Waveform generated.")
 
-    # Save waveform to file
-    torchaudio.save("output.wav", waveform, sample_rate=16000)
+        # Save waveform
+        output_path = "output.wav"
+        torchaudio.save(output_path, waveform, sample_rate=16000)
+        print(f"💾 Audio saved to {output_path}")
 
-    return "output.wav"
+        return output_path
+
+    except Exception as e:
+        print("❌ Error during TTS generation:", e)
+        return "Error during speech synthesis."
 
 # Gradio interface
 demo = gr.Interface(
@@ -47,4 +61,5 @@ demo = gr.Interface(
 )
 
 if __name__ == "__main__":
-    demo.launch()
+    print("🚀 Launching Gradio demo...")
+    demo.launch()