This is rust (using burn) implementation of VoxCPM.

How to use

Build voxcpm-rs.

git clone https://github.com/madushan1000/voxcpm_rs
cd voxcpm_rs
cargo build --release

Download the model weights and convert them.

git clone https://huggingface.co/openbmb/VoxCPM-0.5B
cd voxcpm_rs
cargo run --release --bin voxcpm convert --input-path ../VoxCPM-0.5B/ --output-path burn-models/

Run it.

cargo run --release --bin voxcpm run --model-path burn-models/ \
          --target-text 'VoxCPM is an innovative end-to-end TTS model from ModelBest, designed to generate highly expressive speech.'
mpv output.wav

Or.

cargo run --release --bin voxcpm run --model-path burn-models/ \
            --max-len 200 \
            --target-text 'VoxCPM is an innovative end-to-end TTS model from ModelBest, designed to generate highly expressive speech.' \
            --prompt-text "Having complete focus on a recipe and not allowing yourself to be distracted by your thoughts, can have a therapeutic effect." \
            --prompt-wav-path voices/en_US_joe.wav
mpv output.wav

The example voice is from OHF-Voice public domain voice dataset.