Update app.py

app.py

@@ -9,14 +9,10 @@ f5tts = F5TTS()
 
 @spaces.GPU
 def run_tts(ref_audio, ref_text, gen_text, remove_silence=False):
-    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp_ref:
-        tmp_ref.write(ref_audio.read())
-        tmp_ref_path = tmp_ref.name
-
     output_wav_path = tempfile.mktemp(suffix=".wav")
 
     wav, sr, _ = f5tts.infer(
-        ref_file=tmp_ref_path,
+        ref_file=ref_audio,
         ref_text=ref_text,
         gen_text=gen_text,
         file_wave=output_wav_path,
@@ -28,7 +24,7 @@ def run_tts(ref_audio, ref_text, gen_text, remove_silence=False):
 demo = gr.Interface(
     fn=run_tts,
     inputs=[
-        gr.Audio(label="Reference Audio", type="binary"),
+        gr.Audio(label="Reference Audio", type="filepath"),
         gr.Textbox(label="Reference Text", placeholder="some call me nature, others call me mother nature."),
         gr.Textbox(label="Generation Text", placeholder="I don't really care what you call me..."),
         gr.Checkbox(label="Remove Silence from Output?", value=False)