MyoCodec: A Streaming Neural Codec for Electromyography

Jihwan Lee; Kleanthis Avramidis; Junhyeok Lee; Tiantian Feng; Najim Dehak; Shrikanth Narayanan

eess.SPSep 29, 2026

MyoCodec: A Streaming Neural Codec for Electromyography

Authors: Jihwan Lee, Kleanthis Avramidis, Junhyeok Lee, Tiantian Feng, Najim Dehak, Shrikanth Narayanan

Organizations: Signal Analysis and Interpretation Lab (SAIL), University of Southern California, USA · Center for Language and Speech Processing, Johns Hopkins University, USA

Abstract

Neural codecs encode continuous signals into compact sequences of discrete tokens, providing an interface for efficient transmission, storage, and token-based sequence modeling. This paradigm has been widely adopted in modern speech and audio frameworks; however, the biosignal domain still lacks a neural codec designed specifically for low-bitrate streaming and generalization across diverse downstream tasks. We present MyoCodec, a streaming neural codec designed for electromyography (EMG). Inspired by recent neural audio codecs, MyoCodec combines causal Transformers with residual vector quantization to encode continuous EMG signals into different levels of EMG representations spanning from continuous latent features to discrete tokens operating at 50 Hz. Trained on twelve public EMG datasets, MyoCodec achieves favorable performance in both intrinsic codec quality and representative downstream tasks, including typing (emg2qwerty), hand-pose (emg2pose), speech decoding (emg2speech), and speech-to-EMG synthesis (speech2emg). Across these tasks, MyoCodec exhibits strong performance against prior models while providing a compact and causal EMG representation. During streaming inference, it requires compute time of only 0.482 ms for each 20 ms frame, enabling real-time streaming. Also, the discrete token representation provided by MyoCodec has the potential to support integration into language-model based approaches, creating a path toward LLM-based interactive systems, where tokenized EMG representations are directly processed into such language or speech models. Code and model weights are released.

Figures & tables

Model	Params	Hz	$N_{q}$	bitrate/ch	reconstruction $\uparrow$				codebook
Model	Params	Hz	$N_{q}$	bitrate/ch	SI-SDR	SNR	$r_{\mathrm{wav}}$	$r_{\mathrm{env}}$	util.	perp.
TinyMyo	$3.6\text{\,}\mathrm{M}$	100.0	—	$\leq$ 614k $\dagger$	0.46	1.70	0.735	0.445	—	—
BioCodec	$12.3\text{\,}\mathrm{M}$	27.8	6	1333	$-1.51$	2.20	0.643	0.907	1.000	203.4
MyoCodec (Ours)	$12.7\text{\,}\mathrm{M}$	50.0	2	800	3.62	4.99	0.832	0.933	1.000	172.1
			3	1200	5.89	6.75	0.881	0.949	1.000	177.8
			6	2400	9.80	9.98	0.933	0.967	1.000	186.0

Table 1: Intrinsic codec quality. Reconstruction quality is evaluated using SI-SDR and SNR (dB), waveform Pearson correlation

r_{\mathrm{wav}}

, and envelope Pearson correlation

r_{\mathrm{env}}

N_{q}

is the number of codebooks used. MyoCodec outperforms TinyMyo and BioCodec in signal reconstruction by using only two codebooks (

800\text{\,}\mathrm{b}\mathrm{i}\mathrm{t}\mathrm{s}\mathrm{/}\mathrm{s}

	CB 1	CB 2	CB 3	CB 4	CB 5	CB 6	mean
Utilization	1.000	1.000	1.000	1.000	1.000	1.000	1.000
Perplexity	154.9	189.2	189.2	192.6	194.1	196.0	186.0

Table 2: Per-codebook utilization and perplexity for MyoCodec . All six codebooks are fully utilized, with higher perplexity in later RVQ stages.

EMG Representation	Hz	bitrate/ch	CER (%) $\downarrow$
			greedy		LM beam
			seen	unseen	seen	unseen
Log-spectrogram ( Sivakumar et al., 2024 )	125.0	$\leq$ 132k †	24.02	52.98	16.86	48.57
TinyMyo ( Fasulo et al., 2026 )	100.0	$\leq$ 614k †	21.16	50.02	14.48	44.55
BioCodec ( Avramidis et al., 2025 )	27.8	1.3k	18.52	51.87	13.07	47.78
MyoCodec (Ours)	50.0	2.4k	18.43	50.47	12.74	45.77

Table 3: emg2qwerty. Character error rate (CER) is reported with and without language-model beam search, across seen and unseen subjects. The best and second-best are bolded and underlined , respectively.

EMG encoder	joint-angle error ( ∘ ) $\downarrow$		land. (mm) $\downarrow$	fingertip (mm) $\downarrow$
EMG encoder	val	test	land. (mm) $\downarrow$	fingertip (mm) $\downarrow$
Log-spectrogram ( Salter et al., 2024 )	13.13	15.23	20.76	35.07
TinyMyo ( Fasulo et al., 2026 )	13.15	14.83	20.13	33.97
BioCodec ( Avramidis et al., 2025 )	13.11	14.75	20.12	34.00
MyoCodec (Ours)	12.75	14.24	19.04	32.02

Table 4: emg2pose. Mean joint-angle error (degrees) and mean landmark and fingertip distances (mm).

EMG Encoder	CER $\downarrow$	WER $\downarrow$	PER $\downarrow$	PFER $\downarrow$	UTMOS $\uparrow$
Ground truth	1.43	3.76	7.13	4.19	3.41
Gaddy & Klein ( Gaddy and Klein, 2021 )	22.95	38.49	31.62	14.98	1.90
TinyMyo ( Fasulo et al., 2026 )	16.39	27.77	26.47	13.29	1.91
BioCodec †	16.56	28.33	24.63	12.59	1.93
MyoCodec (Ours)	12.15	21.12	21.61	11.54	2.00

Table 5: emg2speech results. We report character, word, and phoneme error rates (CER, WER, and PER), phone feature error rate (PFER), and UTMOS. All of the downstream models are causally trained with

100\text{\,}\mathrm{m}\mathrm{s}

of lookahead.

		Env. CC $\uparrow$
System	lookahead	$50\text{\,}\mathrm{m}\mathrm{s}$ window	$20\text{\,}\mathrm{m}\mathrm{s}$ window
STE-GAN ( Scheck and Schultz, 2023 )	$\infty$ (offline)	$\mathbf{0.705}$ $\pm$ 0.014	$\mathbf{0.578}$ $\pm$ 0.012
TinyMyo ( Fasulo et al., 2026 )	$\infty$ (offline)	$0.577$ $\pm$ 0.011	$0.474$ $\pm$ 0.010
BioCodec ( Avramidis et al., 2025 )	$0\text{\,}\mathrm{m}\mathrm{s}$	$0.511$ $\pm$ 0.010	$0.434$ $\pm$ 0.008
BioCodec ( Avramidis et al., 2025 )	$144\text{\,}\mathrm{m}\mathrm{s}$	$0.600$ $\pm$ 0.009	$0.522$ $\pm$ 0.008
MyoCodec (Ours)	$0\text{\,}\mathrm{m}\mathrm{s}$	$0.587$ $\pm$ 0.010	$0.503$ $\pm$ 0.009

Table 6: speech2emg. Envelope correlation coefficients (Env. CC) with 95% confidence intervals are reported using

20\text{\,}\mathrm{m}\mathrm{s}

and

50\text{\,}\mathrm{m}\mathrm{s}

smoothing windows. We compare offline and causal systems with different lookahead. The best and second-best are bolded and underlined , respectively.

		Compute time (ms) $\downarrow$			Real-time throughput ( $\times$ ) $\uparrow$
System	basis	Encoder	Decoder	Total	Encoder	Decoder	Total
MyoCodec (Ours)	per second	0.782	0.692	1.474	1279	1445	678
MyoCodec (Ours)	per frame	0.253	0.229	0.482	79	87	41
BioCodec	per second	1.091	0.941	2.032	917	1062	492
BioCodec	per frame	N/A †

Table 7: Inference speed. All measurements are obtained on a single NVIDIA RTX PRO 6000 Blackwell GPU. Per-second speed is measured using

5\text{\,}\mathrm{s}

input windows and normalized by signal duration, while MyoCodec ’s per-frame speed is measured using its compiled CUDA Graph implementation.

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Corpus	ch.	subj.	ch.-hours	content
emg2qwerty ( Sivakumar et al., 2024 )	32	108	11,074	typing
emg2pose ( Salter et al., 2024 )	16	193	5,925	hand pose
Hyser ( Jiang et al., 2021 )	256	20	2,593	finger force
putEMG ( Kaczmarek et al., 2019 )	24	44	1,045	hand gestures
EMG-EPN-612 ( Benalcázar et al., 2020 )	8	612	1,014	hand gestures
MeganePro ( Cognolato et al., 2020 )	12	45	610	grasping

Appendix

Table 8: The twelve public EMG corpora. Channel-hours count only the channels and recordings retained after filtering.

Explore similar work

ZipCodec: Ultra-Low-Frame-Rate Streaming Speech Coding
Sep 12, 2026Luca Della Libera, Cem Subakan, Mirco RavanelliNeural Speech Codecs Dataset Distillation
Concordia University · Mila-Quebec AI Institute · Universit´e Laval
AffectCodec: Emotion-Preserving Neural Speech Codec for Expressive Speech Modeling
May 11, 2026Jiacheng Shi, Hongfei Du, Xinyuan Song +3Neural Speech Codecs Affectcodec
College of William & Mary · Emory University · George Mason University
AffectCodec: Emotion-Preserving Neural Speech Codec with Block-Diagonal Residual FSQ
May 22, 2026Zhaoyang Meng, Zhengyao Ma, Kecan Mao +2Neural Speech Codecs Affectcodec
Beijing University of Posts and Telecommunications

Sep 12, 2026, cs.SDJ/K move · Enter open · S save

ZipCodec: Ultra-Low-Frame-Rate Streaming Speech Coding

Luca Della Libera, Cem Subakan, Mirco Ravanelli

Concordia University · Mila-Quebec AI Institute · Universit´e Laval

Neural Speech Codecs Dataset Distillation

Open paper AskPDF arXiv

Neural audio codecs are a fundamental component of modern speech generation systems. While recent codecs achieve increasingly low bitrates, reducing frame rate remains challenging, as each token must preserve more information while maintaining reconstruction quality. We present ZipCodec, a streaming neural speech codec operating at 6.25 Hz and 0.80 kbps with a theoretical latency of 160 ms. Our approach combines large-scale WavLM distillation with a redesigned transformer-based architecture, a scalar spherical quantizer, and a latency-aware streaming decoder. Experiments show that ZipCodec substantially outperforms existing streaming codecs at comparable bitrates in both reconstruction and downstream tasks, while operating at a significantly lower frame rate. Despite its 842M parameters, ZipCodec achieves real-time single-stream inference on a consumer-grade CPU. Demo samples, code and checkpoints are available at https://lucadellalib.github.io/zipcodec-web/.

Abstract

Explore similar work

ZipCodec: Ultra-Low-Frame-Rate Streaming Speech Coding

AffectCodec: Emotion-Preserving Neural Speech Codec for Expressive Speech Modeling

AffectCodec: Emotion-Preserving Neural Speech Codec with Block-Diagonal Residual FSQ