Update app.py

app.py

@@ -75,7 +75,7 @@ def inference(
 
     start_time = time.time()
 
-    # all_parts = []
+    all_parts = []
     for j, text in enumerate(texts):
         for audio_frame in tts.tts_with_preset(
             text,
@@ -84,14 +84,14 @@ def inference(
             preset=preset,
             k=1
         ):
-            # print("Time taken: ", time.time() - start_time)
-            # all_parts.append(audio_frame)
-            yield (24000, audio_frame.cpu().detach().numpy())
+            all_parts.append(audio_frame)
+
+    wav = torch.cat(all_parts, dim=0).unsqueeze(0)
+    output_file = "output.wav"
+    torchaudio.save(output_file, wav.cpu(), 24000)
+
+    return output_file
 
-    # wav = torch.cat(all_parts, dim=0).unsqueeze(0)
-    # print(wav.shape)
-    # torchaudio.save("output.wav", wav.cpu(), 24000)
-    # yield (None, gr.make_waveform(audio="output.wav",))
 
 def main():
     title = "Tortoise TTS 🐢"
@@ -133,8 +133,7 @@ def main():
         value="No",
     )
 
-    output_audio = gr.Audio(label="streaming audio:", streaming=True, autoplay=True)
-    # download_audio = gr.Audio(label="dowanload audio:")
+    download_audio = gr.Audio(label="Download audio:", download=True)
     interface = gr.Interface(
         fn=inference,
         inputs=[
@@ -147,9 +146,9 @@ def main():
         ],
         title=title,
         description=description,
-        outputs=[output_audio],
+        outputs=[download_audio],
     )
-    interface.queue().launch()
+    interface.launch()
 
 
 if __name__ == "__main__":