mygyasir's picture
Duplicate from konverner/deep-voice-cloning
13c43fe
import os
from pathlib import Path
import gradio as gr
os.system('pip install .')
def greet(text, audio_file_path, progress=gr.Progress()):
text = "%s" % text
audio_file_path = "%s" % audio_file_path
out_path = Path("scripts/output/audio.wav")
progress(0.2, desc="Training voice embedding... (aprx 20 mins)")
os.system(f'python scripts/train.py --audio_path {audio_file_path}\
--output_dir "models"')
progress(0.9, desc="Generating voice...")
os.system(f'python scripts/cloning_inference.py --model_path "models/microsoft_speecht5_tts_{Path(audio_file_path).stem}"\
--input_text "{text}" --output_path "{str(out_path)}"')
return out_path
demo = gr.Interface(
fn=greet,
inputs=[gr.Textbox(label='What would you like the voice to say? (max. 2000 characters per request)'),
gr.Audio(type="filepath", source="upload", label='Upload a voice to clone (max. 50mb)')],
outputs="audio",
title="Deep Voice Cloning Tool"
)
demo.launch()