Update app.py

app.py

@@ -76,9 +76,8 @@ def whisper_speech_demo(multilingual_text, speaker_audio):
     # Debug statement        print("Final concatenated audio shape:", concatenated_audio.shape)
     concatenated_audio = concatenated_audio / np.max(np.abs(concatenated_audio))
 
-    with tempfile.NamedTemporaryFile(suffix='.wav', delete=False) as tmp_file:
-        sf.write(tmp_file.name, concatenated_audio.T, 24000, format='WAV', subtype='PCM_16')
-        return tmp_file.name
+    return (24000, concatenated_audio.T)
+
 
 with gr.Blocks() as demo:
     gr.Markdown(title)