# NeuroRVQ: Multi-Scale Biosignal Tokenization for Generative Foundation Models

arXiv: https://arxiv.org/abs/2510.13068 · HTML: https://mlsanity.com/paper/2510.13068/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 16 of its 63 references that are in the map.
- First posted on arXiv 15 Oct 2025.

## Builds on (most cited first, 16 of 16)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [Taming Transformers for High-Resolution Image Synthesis](https://mlsanity.com/paper/2012.09841/) (2020)
- [High Fidelity Neural Audio Compression](https://mlsanity.com/paper/2210.13438/) (2022)
- [EEGNet: A Compact Convolutional Network for EEG-based Brain-Computer Interfaces](https://mlsanity.com/paper/1611.08024/) (2016)
- [Large Brain Model for Learning Generic Representations with Tremendous EEG Data in BCI](https://mlsanity.com/paper/2405.18765/) (2024)
- [CBraMod: A Criss-Cross Brain Foundation Model for EEG Decoding](https://mlsanity.com/paper/2412.07236/) (2024)
- [Automatic diagnosis of the 12-lead ECG using a deep neural network](https://mlsanity.com/paper/1904.01949/) (2019)
- [4M: Massively Multimodal Masked Modeling](https://mlsanity.com/paper/2312.06647/) (2023)
- [The Temple University Hospital Seizure Detection Corpus](https://mlsanity.com/paper/1801.08085/) (2018)
- [Deep Learning for Electromyographic Hand Gesture Signal Classification Using Transfer Learning](https://mlsanity.com/paper/1801.07756/) (2018)
- [BrainOmni: A Brain Foundation Model for Unified EEG and MEG Signals](https://mlsanity.com/paper/2505.18185/) (2025)
- [emg2pose: A Large and Diverse Benchmark for Surface Electromyographic Hand Pose Estimation](https://mlsanity.com/paper/2412.02725/) (2024)
- [emg2qwerty: A Large Dataset with Baselines for Touch Typing using Surface Electromyography](https://mlsanity.com/paper/2410.20081/) (2024)
- [Prediction of Reaction Time and Vigilance Variability from Spatiospectral Features of Resting-State EEG in a Long Sustained Attention Task](https://mlsanity.com/paper/1910.10076/) (2019)
- [PhysioWave: A Multi-Scale Wavelet-Transformer for Physiological Signal Representation](https://mlsanity.com/paper/2506.10351/) (2025)

## Built on by (most cited first, 6 of 6)

- [NeuroAtlas: Benchmarking Foundation Models for Clinical EEG and Brain-Computer Interfaces](https://mlsanity.com/paper/2605.14698/) (2026)
- [A Systematic Evaluation of Sample-Level Tokenization Strategies for MEG Foundation Models](https://mlsanity.com/paper/2602.16626/) (2026)
- [Scaling Next-Brain-Token Prediction for MEG](https://mlsanity.com/paper/2601.20138/) (2026)
- [EMGBlend: Heterogeneity-Aware Self-Supervised Pretraining for Gesture and Force Decoding](https://mlsanity.com/paper/2609.25582/) (2026)
- [Benchmarking EEG Foundation Models at Scale: Lessons from 20,000 Evaluations](https://mlsanity.com/paper/2609.32743/) (2026)
- [EMG-GPT: Predictive Pretraining on Residual-Quantized EMG Tokens for Hand Pose Estimation](https://mlsanity.com/paper/2610.05235/) (2026)

## Abstract (arXiv)

Biosignals such as electroencephalography (EEG), electrocardiography (ECG), and electromyography (EMG) encode physiological activity across multiple temporal and spectral scales, yielding representations that are rich but challenging for machine learning. Foundation models trained to predict masked signal tokens have shown promise in learning generalizable biosignal representations, yet their performance depends on the tokenizer's ability to preserve high-frequency dynamics and reconstruct signals with high fidelity. We introduce NeuroRVQ, a modality-adaptive biosignal tokenizer family designed for high-fidelity signal reconstruction. To capture the full frequency spectrum, NeuroRVQ decomposes biosignals into frequency-specific representations via multi-scale temporal convolutions, each encoded into hierarchical RVQ codebooks to preserve high-frequency detail, combined with a novel phase-aware training loss that respects the circular topology of Fourier phase. By tuning the temporal resolution, number and size of temporal kernels and RVQ depth, this design adapts to the spectro-temporal characteristics of each biosignal modality. To validate that tokenizer quality drives downstream performance, we train a simple masked-token foundation model for each modality (NeuroRVQ-FM) using the corresponding NeuroRVQ tokenizer. The NeuroRVQ-FM family achieves competitive or superior downstream performance compared to existing modality-specific foundation models, demonstrating that high-fidelity tokenization is a critical factor for effective biosignal modeling.