Spaces:
Sleeping
Sleeping
| import gradio as gr | |
| import edge_tts | |
| import asyncio | |
| import tempfile | |
| # 1) Load voices once at startup | |
| def load_voices(): | |
| loop = asyncio.get_event_loop() | |
| voices = loop.run_until_complete(edge_tts.list_voices()) | |
| return { | |
| f"{v['ShortName']} - {v['Locale']} ({v['Gender']})": v['ShortName'] | |
| for v in voices | |
| } | |
| VOICES = load_voices() | |
| # 2) Async TTS worker | |
| async def _text_to_speech(text, short_name, rate_str, pitch_str): | |
| comm = edge_tts.Communicate(text, short_name, rate=rate_str, pitch=pitch_str) | |
| with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmp: | |
| await comm.save(tmp.name) | |
| return tmp.name | |
| # 3) Sync wrapper for Gradio callback | |
| def tts_interface(text, voice_choice, rate, pitch): | |
| if not text.strip(): | |
| return None, "🚨 Please enter some text." | |
| if not voice_choice: | |
| return None, "🚨 Please select a voice." | |
| short_name = voice_choice.split(" - ")[0] | |
| rate_str = f"{rate:+d}%" | |
| pitch_str = f"{pitch:+d}Hz" | |
| try: | |
| audio_path = asyncio.get_event_loop().run_until_complete( | |
| _text_to_speech(text, short_name, rate_str, pitch_str) | |
| ) | |
| return audio_path, "" | |
| except Exception as e: | |
| return None, f"❌ TTS failed: {e}" | |
| # 4) Build the UI | |
| def create_demo(): | |
| with gr.Blocks(analytics_enabled=False) as demo: | |
| gr.Markdown("# 🎙️ Edge TTS on Hugging Face Spaces") | |
| gr.Markdown( | |
| "**Convert your text to speech** using Microsoft Edge’s neural voices. " | |
| "Adjust rate and pitch to fine-tune the output." | |
| ) | |
| with gr.Row(): | |
| txt = gr.Textbox(label="Input Text", lines=5, placeholder="Type something…") | |
| vox = gr.Dropdown(choices=list(VOICES.keys()), label="Voice") | |
| rate = gr.Slider(-50, 50, value=0, label="Rate (%)") | |
| pitch = gr.Slider(-20, 20, value=0, label="Pitch (Hz)") | |
| btn = gr.Button("Generate Speech") | |
| audio_out = gr.Audio(type="filepath", label="Audio Output") | |
| warn_md = gr.Markdown("", label="Warnings / Errors") | |
| btn.click( | |
| fn=tts_interface, | |
| inputs=[txt, vox, rate, pitch], | |
| outputs=[audio_out, warn_md] | |
| ) | |
| # Enable the queue so the API actually gets registered | |
| demo.queue() | |
| return demo | |
| # 5) Launch with SSR disabled | |
| if __name__ == "__main__": | |
| demo = create_demo() | |
| demo.launch(ssr=False) | |