update tts

app.py

@@ -25,18 +25,18 @@ def generate_tts(text: str, speaker: int = 0):
     spectrogram = spec_generator.generate_spectrogram(tokens=parsed, speaker=speaker)
     audio = voc_model.convert_spectrogram_to_audio(spec=spectrogram)
 
-    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
-        torchaudio.save(fp.name, audio.to('cpu'), sample_rate=sr)
+    # with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
+    #     torchaudio.save(fp.name, audio.to('cpu'), sample_rate=sr)
 
-        return fp.name
-    #return (sr, audio.to('cpu').detach().numpy())
+    #     return fp.name
+    return (sr, audio.to('cpu').detach().numpy())
 
 def run():
     demo = gr.Interface(
         fn=generate_tts,
         inputs=[gr.Textbox(value="This is a test.", label="Text to Synthesize"),
                 gr.Slider(0, 10, step=1, label="Speaker")],
-        outputs=gr.Audio(label="Output", type="filepath"),
+        outputs=gr.Audio(label="Output", type="numpy"),
     )
 
     demo.launch(server_name="0.0.0.0", server_port=7860)