anzeyimana/KinyaTTS

A codebase for Kinyarwanda speech synthesis (text-to-speech) based on MB-iSTFT-VITS2 model in PyTorch # KinyaTTS A codebase for Kinyarwanda speech synthesis (text-to-speech) based on MB-iSTFT-VITS2 model in PyTorch. The original codebase and description comes from MB-iSTFT-VITS2 which itself is a hybrid combination of vits2_pytorch and MB-iSTFT-VITS. An architectural depiction of the model is presented below and its details can be found the original sources. ## Getting started ### Inference 2. Checkout the code in the Inference directory and install `monotonic_align` the `kinyatts` modules ````sh pip install -e ./Inference/monotonic_align/ pip install -e ./Inference/ ```` 3. Download the pre-trained Kinyarwanda TTS model TTS_MODEL_ms_ktjw_istft_vits2_base_1M.pt 4. Go to the `kinyatts` sub-directory and run an inference server using uwsgi ```` cd Inference/kinyatts/ nohup sh run.sh & ```` 5. Alternatively use the provided Jupiter notebook to synthetise speech: Inference/kinyatts/kinyatts_inference.ipynb ### Training Follow instructions in Training codebase to install requirements and train a basic multi-speaker TTS model. ## Credits - FENRlR/MB-iSTFT-VITS2 - jaywalnut310/vits - p0p4k/vits2_pytorch - MasayaKawamura/MB-iSTFT-VITS - ORI-Muchim/PolyLangVITS - tonnetonne814/MB-iSTFT-VITS-44100-Ja - misakiudon/MB-iSTFT-VITS-multilingual