BioCodec (USC SAIL; neural-codec biosignal tokenizer)
- Year
- 2025
- Form factor
- Other
Activity counts measure attention, not model quality. Missing counts are unknown. Publication status follows the linked source; a listed venue alone does not confirm peer review.
BioCodec is an EnCodec-style neural codec that tokenizes single-channel biosignals with residual vector quantization. It was pretrained on large-scale EEG and evaluated on clinical diagnosis, sleep, speech, and motor-imagery decoding, with secondary EMG tests. MIT-licensed code and an in-repo checkpoint are available (arXiv 2510.09095).
- Signals
- EEG, EMG
- Pretraining data
- Thousands of EEG hours (per abstract)
- Form factor
- Other
- Year
- 2025
- Architecture
EnCodec-style convolutional neural audio codec adapted to single-channel biosignals: encoder → residual vector quantizer (K=6 stages, vocabulary 256, code dim 16; per BrainCodes §2.1) → decoder; produces discrete tokens per channel
- License
MIT (repo LICENSE); the in-repo checkpoint is covered by the same licence (no separate weights terms).
- Type
- Foundation model
- Form factor
- Other
- Parameters
- Not reported
- Architecture
- EnCodec-style convolutional neural audio codec adapted to single-channel biosignals: encoder → residual vector quantizer (K=6 stages, vocabulary 256, code dim 16; per BrainCodes §2.1) → decoder; produces discrete tokens per channel
- Venue
- arXiv 2510.09095v1, 2025-10-10 ('currently under peer review')
- Year
- 2025
- DOI
- Not reported
- Weights
- Open
- Licence
- MIT (repo LICENSE); the in-repo checkpoint is covered by the same licence (no separate weights terms).
- Caveats
- Not reported
- Datasets
- Not reported
- Added
- Oct 22, 2026
- Updated
- Oct 26, 2026
Abstract
Neurophysiological recordings such as electroencephalography (EEG) offer accessible and minimally invasive means of estimating physiological activity for applications in healthcare, diagnostic screening, and even immersive entertainment. However, these recordings yield high-dimensional, noisy time-series data that typically require extensive pre-processing and handcrafted feature extraction to reveal meaningful information. Recently, there has been a surge of interest in applying representation learning techniques from large pre-trained (foundation) models to effectively decode and interpret biosignals. We discuss the challenges posed for incorporating such methods and introduce BioCodec, an alternative representation learning framework inspired by neural codecs to capture low-level signal characteristics in the form of discrete tokens. Pre-trained on thousands of EEG hours, BioCodec shows efficacy across multiple downstream tasks, ranging from clinical diagnostic tasks and sleep physiology to decoding speech and motor imagery, particularly in low-resource settings. Additionally, we provide a qualitative analysis of codebook usage and estimate the spatial coherence of codebook embeddings from EEG connectivity. Notably, we also document the suitability of our method to other biosignal data, i.e., electromyographic (EMG) signals. Overall, the proposed approach provides a versatile solution for biosignal tokenization that performs competitively with state-of-the-art models. The source code and model checkpoints are shared.
Verbatim fromarXiv· arXiv metadata (abstracts are released under CC0)
How it works
- Pretraining data
- Thousands of EEG hours (per abstract)
- Architecture
- EnCodec-style convolutional neural audio codec adapted to single-channel biosignals: encoder → residual vector quantizer (K=6 stages, vocabulary 256, code dim 16; per BrainCodes §2.1) → decoder; produces discrete tokens per channel
- Inputs
- EEG, EMG
- Approach
- tokenizer · neural codec · RVQ · EEG · EMG · open weights
Datasets
Cite
APA
BioCodec (USC SAIL; neural-codec biosignal tokenizer). (2025). arXiv 2510.09095v1, 2025-10-10. https://arxiv.org/abs/2510.09095
BibTeX
@misc{biocodec2025biocodec,
title = {{BioCodec (USC SAIL; neural-codec biosignal tokenizer)}},
year = {2025},
howpublished = {arXiv 2510.09095v1, 2025-10-10},
eprint = {2510.09095},
archivePrefix = {arXiv}
}Built from the hub record. Check it against the publisher before submitting.
History
Record updated
Added to the ChatPPG research hub
