Update app.py

app.py

@@ -1,6 +1,40 @@
 import gradio as gr
+from f5_tts.infer.utils_infer import remove_silence_for_generated_wav
+from demo_f5tts import F5TTS  # assuming your F5TTS class is saved in demo_f5tts.py
+import tempfile
+import os
 
-with gr.Blocks() as demo:
-    gr.Markdown("Hi everyone, due to breaking changes with ZeroGPU/Xet-storage spaces, this space is temporarily down. I hope to find a solution to this soon, so please stay tuned. Sorry for the inconvenience. In the mean time, please check out: https://huggingface.co/spaces/mrfakename/MegaTTS3-Voice-Cloning https://huggingface.co/spaces/styletts2/styletts2 if you need TTS spaces.")
+f5tts = F5TTS()
 
-demo.launch()
+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_text=ref_text,
+        gen_text=gen_text,
+        file_wave=output_wav_path,
+        remove_silence=remove_silence,
+    )
+
+    return output_wav_path
+
+demo = gr.Interface(
+    fn=run_tts,
+    inputs=[
+        gr.Audio(label="Reference Audio", type="binary"),
+        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)
+    ],
+    outputs=gr.Audio(label="Generated Speech"),
+    title="🗣️ F5-TTS Demo",
+    description="Upload a reference voice, give reference and generation text, and hear it in the same voice!",
+)
+
+if __name__ == "__main__":
+    demo.launch()