Update app.py

app.py

@@ -78,7 +78,7 @@ processor = SpeechT5Processor.from_pretrained(checkpoint)
 
 
 def generate_wav_file(text):
-  try:  
+    try:  
       converted_text = replace_numbers_with_words(text)
       cleaned_text = cleanup_text(converted_text)
       final_text = normalize_text(cleaned_text)
@@ -86,16 +86,18 @@ def generate_wav_file(text):
       speaker_embeddings = torch.tensor(np.load('speaker_embedding.npy'))
       vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan")
       speech = model.generate_speech(inputs["input_ids"], speaker_embeddings, vocoder=vocoder)
-      return Audio(speech.numpy(), rate=16000)
-  except Exception as e:
+    # Convert the speech to a WAV file
+      output_file = "output.wav"
+      sf.write(output_file, speech.detach().cpu().numpy(), 16000)
+      return output_file  # Return the file path for download
+    except Exception as e:
         print(f"Error: {e}")
-        return None 
-      
+        return None
 
 iface = gr.Interface(
     fn=generate_wav_file,
     inputs=gr.Textbox(lines=3, label="Enter text to convert to speech"),
-    outputs="audio",
+    outputs=gr.Audio(type="file", label="Generated Audio"),
     title="Text-to-Speech Technical EN"
 )
 if __name__ == "__main__":