Skip to content

Latest commit

 

History

History
71 lines (46 loc) · 2.29 KB

README.md

File metadata and controls

71 lines (46 loc) · 2.29 KB

SYLBER: Syllabic Embedding Representation of Speech from Raw Audio

Paper | Audio Samples

Updates

12/25/2024

  1. Initial code release with training and inference pipelines.
  2. Checkpoint release

Environment

Install the dependencies from requirements.txt:

pip install -r requirements.txt

Training SYLBER

Datasets and Checkpoints

  1. Noise Dataset for WavLM-based Augmentation: The noise dataset for the WavLM noise augmentation is sourced from DNS Challenge. You can use the following script to download the dataset:

    bash download-dns-challenge-3.sh

    and untar datasets_fullband/datasets_fullband.noise_fullband.tar.bz2

  2. Generated Datasets: The other data used for training SYLBER are generated using the SDHuBERT repository. Please follow the instructions there for data preparation.

  3. Checkpoints: Pretrained model checkpoints for sylber are available on Google Drive: linik

Stage 1 Training

python train.py --config-name=sylber_base

Stage 2 Training

python train.py --config-name=sylber_base_stage2

The training is split into two stages. Make sure to review the configurations in the configs/ directory for detailed settings.

Inference

Segmentation and Visualization

For inference to obtain segmentations and visualize results, please refer to demo.ipynb.

SPARC (formerly known as Articulatory Encodec)

For using SPARC, refer to Speech-Articulatory-Coding for installation and usage instructions.

Acknowledgements

Website adapted from: https://github.com/BytedanceSpeech/bytedancespeech.github.io

Citation

If you use this work, please cite our paper:

@article{cho2024sylber,
  title={Sylber: Syllabic Embedding Representation of Speech from Raw Audio},
  author={Cho, Cheol Jun and Lee, Nicholas and Gupta, Akshat and Agarwal, Dhruv and Chen, Ethan and Black, Alan W and Anumanchipalli, Gopala K},
  journal={arXiv preprint arXiv:2410.07168},
  year={2024}
}