Barkify: an unoffical repo for training Bark, a text-prompted generative audio model by suno-ai.

Bark has two GPT style models which is compatible for prompting and other tricks from NLP. Bark realize a great real world tts result but the repo itself doesn't a training recipe. We want to conduct some experiments or train this model. Here we release our basic training code which might be a guidance of training for open source community.

We do our experiment on LJspeech. Follow the instrcutions in process.ipynb.
For chinese, we test a famous steamer named Fengge. It shows an acceptable result but worse than our other TTS repo.

Stage1 stands for text to semantic and stage2 stands for semantic to acoustic.
You should config paramters in the configs/barkify.yaml. We use one A100 to train our model (both S1&S2).

# training stage 1 or 2
python trainer.py start_path=/path/to/your/work_env stage=1 name=<dataset>
python trainer.py start_path=/path/to/your/work_env stage=2 name=<dataset>

Directly use infer.ipynb and follow the instrcutions to infer your model.

• Construct a basic training code for bark-like generative model
• Test one speaker scenario
• Test multi speaker scenario
• Test speaker semantic prompting
• Test speech/audio acoustic prompting
• Test variable length data(as we use a fixed length now)
• Long-form generation
• Support more language

• bark is a transformer-based text-to-audio model.
• Vall-E is an unofficial PyTorch implementation of VALL-E.