about
GSLM2 MultiStream Speech Rep Learning and Gen Spoken Language Modeling (arxiv.org)
3 points by mrkn1 14 days ago | hide | past | pdf | discuss on HN

In plain words: Instead of squeezing speech into a single stream of word-like units, this system learns several layers of units capturing different speech details and predicts them together. It cuts the gap to text-trained language models by about 100-fold on grammar and word tests.

Abstract · GSLM2: Multi-Stream Speech Representation Learning and Generative Spoken Language Modeling

Generative spoken language models (SLMs) offer a route to learning linguistic structure purely from speech, yet they lag far behind text-based models in both capability and data scaling efficiency. We identify two bottlenecks underlying this gap: the limited linguistic quality of existing speech units and the reliance on a single stream of discrete units to represent speech. Multiple codebooks are widely used to encode acoustic features but rarely to represent speech semantics. We present GSLM2, which learns and models complementary properties of speech at multiple levels of abstraction. Its encoder, SpidR 2.0, uses self-supervised learning with differentiable prediction heads to induce high-quality discrete units at multiple levels. A multi-stream SLM then jointly models these units using a shared temporal backbone, stream-specific predictors, and an autoregressive head that captures cross-stream dependencies. Across training data sizes and model scales, up to 6M hours and 6.5B parameters, GSLM2 consistently improves linguistic competence while scaling more efficiently with data. As a result, it substantially narrows the gap between spoken and text-based LMs; across read and conversational speech domains, we shrink the gap by about two orders of magnitude on sWuggy, sBLIMP, and tSC, while also outperforming competitive speech-text systems on all three tasks. We open-source the training code and model checkpoints at https://github.com/facebookresearch/gslm2.

Yosuke Higuchi, Tom Fizycki, Mahi Luthra, Charles-Éric Saint-James, Azeez Syed, Angel Villar-Corrales, Mido Assran, Nicolas Ballas, Jiayi Shen, Dongyan Lin, Phillip Rust, Surya Parimi, et al.
arXiv:2611.01143 · cs.CL, cs.SD, eess.AS · submitted Oct 31, 2026
abstract · pdf · html · Equal contribution by Yosuke Higuchi and Tom Fizycki. Core contribution by Mahi Luthra, Charles-Éric Saint-James and Azeez Syed

add comment on HN