From Hallucination to Articulation: Language Model-Driven Losses for Ultra Low-Bitrate Neural Speech Coding
Abstract
``Phoneme Hallucinations (PH)'' commonly occur in low-bitrate DNN-based codecs. It is the generative decoder's attempt to synthesize plausible outputs from excessively compressed tokens missing some semantic information. In this work, we propose language model-driven losses (LM loss) and show they may alleviate PHs better than a semantic distillation (SD) objective in very-low-bitrate settings. The proposed LM losses build upon language models pretrained to associate speech with text. When ground-truth transcripts are unavailable, we propose to modify a popular automatic speech recognition (ASR) model, Whisper, to compare the decoded utterance against the ASR-inferred transcriptions of the input speech. Else, we propose to use the timed-text regularizer (TTR) to compare WavLM representations of the decoded utterance against BERT representations of the ground-truth transcriptions. We test and compare LM losses against an SD objective, using a reference codec whose three-stage training regimen was designed after several popular codecs. Subjective and objective evaluations conclude that LM losses may provide stronger guidance to extract semantic information from self-supervised speech representations, boosting human-perceived semantic adherence while preserving overall output quality. Demo samples, code, and checkpoints are available online.
Cite
@article{arxiv.2602.06213,
title = {From Hallucination to Articulation: Language Model-Driven Losses for Ultra Low-Bitrate Neural Speech Coding},
author = {Jayeon Yi and Minje Kim},
journal= {arXiv preprint arXiv:2602.06213},
year = {2026}
}
Comments
To appear in ICASSP 2026. Demo wavs, code, and checkpoints (currently) availble at https://github.com/stet-stet/lmloss-icassp2026