import soundfile as sf
from voxcpm import VoxCPM
model = VoxCPM.from_pretrained("ghananlpcommunity/ghana-tts-72k")
wav = model.generate(
text="VoxCPM is an innovative end-to-end TTS model from ModelBest, designed to generate highly expressive speech.",
prompt_wav_path=None, # optional: path to a prompt speech for voice cloning
prompt_text=None, # optional: reference text
cfg_value=2.0, # LM guidance on LocDiT, higher for better adherence to the prompt, but maybe worse
inference_timesteps=10, # LocDiT inference timesteps, higher for better result, lower for fast speed
normalize=True, # enable external TN tool
denoise=True, # enable external Denoise tool
retry_badcase=True, # enable retrying mode for some bad cases (unstoppable)
retry_badcase_max_times=3, # maximum retrying times
retry_badcase_ratio_threshold=6.0, # maximum length restriction for bad case detection (simple but effective), it could be adjusted for slow pace speech
)
sf.write("output.wav", wav, 16000)
print("saved: output.wav")Ghana TTS – 72k
Checkpoint at 72,000 steps for the Ghana TTS model, fine-tuned from VoxCPM on 1,400+ hours of Ghanaian language speech data across 44 languages.
Trained on a single NVIDIA H200 (140 GB VRAM) using bf16 + TF32.
Contents
model.safetensors– model weightsoptimizer.pth– optimizer stateaudiovae/– audio VAE componentsscheduler/– noise scheduler configtokenizer/– tokenizer filesprompt_audio/– reference audio samples for voice prompting
Usage
See the Ghana TTS space for inference.
Training
Training code: github.com/michsethowusu/VoxCPM
- Downloads last month
- -
# Gated model: Login with a HF token with gated access permission hf auth login