xtts / app.py
Vaibhav Srivastav
up
142fdc7
raw
history blame
1.45 kB
import gradio as gr
from TTS.api import TTS
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v1", gpu=True)
def predict(prompt, language, audio_file_pth):
tts.tts_to_file(
text=prompt,
file_path="output.wav",
speaker_wav=audio_file_pth,
language=language,
)
return gr.make_waveform(
audio="output.wav",
)
title = "XTTS: MVP"
gr.Interface(
fn=predict,
inputs=[
gr.Textbox(
label="Text Prompt",
info="One or two sentences at a time is better",
placeholder="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",
],
max_choices=1,
),
gr.Audio(
label="Reference Audio",
info="Upload a reference audio for target speaker voice",
type="filepath",
),
],
outputs=[
gr.Video(label="Synthesised Speech"),
],
title=title,
).launch(debug=True)