Update app.py

app.py

@@ -11,6 +11,7 @@ subprocess.check_call([sys.executable, "-m", "pip", "install", "entmax"])
 from tortoise_tts.api import TextToSpeech
 from tortoise_tts.utils.audio import load_audio, get_voices
 import torch 
+import torchaudio
 import gradio as gr
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
@@ -47,7 +48,9 @@ def inference(text, voice):
         conds.append(c)
     print(text, conds, preset)
     gen = tts.tts_with_preset(text, conds, preset)
-    return gen
+    print("gen")
+    torchaudio.save('generated.wav', gen.squeeze(0).cpu(), 24000)
+    return "generated.wav"
  
 text = "Joining two modalities results in a surprising increase in generalization! What would happen if we combined them all?"
 iface = gr.Interface(
@@ -56,7 +59,7 @@ iface = gr.Interface(
       gr.inputs.Textbox(type="str", default=text, label="Text", lines=3),
       gr.inputs.Dropdown(voices),
   ],
-  outputs="audio",
+  outputs="text",
   title="TorToiSe",
   description="A multi-voice TTS system trained with an emphasis on quality",
   article="This demo shows the ultra fast option in the TorToiSe system. For more info check the <a href='https://github.com/neonbjb/tortoise-tts' target='_blank'>Repository</a>.",