import gradio as gr import sys from TTS.api import TTS tts = TTS("tts_models/multilingual/multi-dataset/xtts_v1") tts.to("cuda") def predict(prompt, language, audio_file_pth, mic_file_path, use_mic, agree): if agree == True: if use_mic == True: if mic_file_path is not None: speaker_wav=mic_file_path else: gr.Warning("Please record your voice with Microphone, or uncheck Use Microphone to use reference audios") return ( None, None, ) else: speaker_wav=audio_file_pth if len(prompt)<2: gr.Warning("Please give a longer prompt text") return ( None, None, ) try: tts.tts_to_file( text=prompt, file_path="output.wav", speaker_wav=speaker_wav, language=language, ) except RuntimeError: if "device-side" in e.message: # cannot do anything on cuda device side error, need tor estart gr.Warning("Unhandled Exception encounter, please retry in a minute") print("Cuda device-assert Runtime encountered need restart") sys.exit("Exit due to cuda device-assert") raise return ( gr.make_waveform( audio="output.wav", ), "output.wav", ) else: gr.Warning("Please accept the Terms & Condition!") return ( None, None, ) title = "Coqui🐸 XTTS" description = """ XTTS is a Voice generation model that lets you clone voices into different languages by using just a quick 3-second audio clip.
Built on Tortoise, XTTS has important model changes that make cross-language voice cloning and multi-lingual speech generation super easy.
This is the same model that powers Coqui Studio, and Coqui API, however we apply a few tricks to make it faster and support streaming inference.

For faster inference without waiting in the queue, you should duplicate this space and upgrade to GPU via the settings.
Duplicate Space

""" article = """

By using this demo you agree to the terms of the Coqui Public Model License at https://coqui.ai/cpml

""" examples = [ [ "Once when I was six years old I saw a magnificent picture", "en", "examples/female.wav", None, False, True, ], [ "Lorsque j'avais six ans j'ai vu, une fois, une magnifique image", "fr", "examples/male.wav", None, False, True, ], [ "Un tempo lontano, quando avevo sei anni, vidi un magnifico disegno", "it", "examples/female.wav", None, False, True, ], [ "Bir zamanlar, altı yaşındayken, muhteşem bir resim gördüm", "tr", "examples/female.wav", None, False, True, ], ] gr.Interface( fn=predict, inputs=[ gr.Textbox( label="Text Prompt", info="One or two sentences at a time is better", value="It took me quite a long time to develop a voice, and now that I have it I'm not going to be silent.", ), gr.Dropdown( label="Language", info="Select an output language for the synthesised speech", choices=[ "en", "es", "fr", "de", "it", "pt", "pl", "tr", "ru", "nl", "cz", "ar", "zh-cn", ], max_choices=1, value="en", ), gr.Audio( label="Reference Audio", info="Click on the ✎ button to upload your own target speaker audio", type="filepath", value="examples/female.wav", ), gr.Audio(source="microphone", type="filepath", info="Use your microphone to record audio", label="Use Microphone for Reference"), gr.Checkbox(label="Check to use Microphone as Reference", value=False), gr.Checkbox( label="Agree", value=False, info="I agree to the terms of the Coqui Public Model License at https://coqui.ai/cpml", ), ], outputs=[ gr.Video(label="Waveform Visual"), gr.Audio(label="Synthesised Audio"), ], title=title, description=description, article=article, examples=examples, ).queue().launch(debug=True)