My VoxCPM2 TTS Model
This model is based on openbmb/VoxCPM2, a multilingual Text-to-Speech model released under the Apache-2.0 license.
This repository contains my duplicated and/or fine-tuned version for building a custom TTS service or application.
Features
- Multilingual text-to-speech
- Voice design from text description
- Voice cloning from reference audio
- 48kHz audio output
- Suitable for personal and commercial applications, subject to the Apache-2.0 license and responsible-use restrictions
Usage
from voxcpm import VoxCPM
import soundfile as sf
model = VoxCPM.from_pretrained("ravuthz/VoxCPM2-khmer", load_denoiser=False)
wav = model.generate(
text="VoxCPM2 គឺជាការចេញផ្សាយដែលបានណែនាំបច្ចុប្បន្នសម្រាប់ការសំយោគការនិយាយច្រើនភាសាដែលប្រាកដនិយម។",
cfg_value=2.0,
inference_timesteps=10,
)
sf.write("output.wav", wav, model.tts_model.sample_rate)