Whisper-GPT: Hybrid Continuous-Discrete Model for Speech and Music Generation
Researchers have proposed WHISPER-GPT, a generative large language model that simultaneously processes continuous audio representations (spectrograms) and discrete acoustic tokens for speech and music generation. The work addresses a key limitation of existing discrete-token-only audio models, namely the explosion in context length required for high-fidelity generation. The hybrid approach improves perplexity and negative log-likelihood scores for next-token prediction, suggesting more efficient and accurate audio modeling.
WHISPER-GPT is a proposed generative language model architecture designed to handle both continuous audio representations, such as spectrograms, and discrete acoustic tokens derived from neural compression algorithms like EnCodec within a single unified framework. Current state-of-the-art generative audio models predominantly rely on discrete tokens, but this approach suffers from rapidly expanding context lengths when high-fidelity audio across multiple frequency bands must be modeled. By incorporating spectrograms, the architecture encodes all relevant audio information at a given time instant into a single token, while still leveraging the discrete space for sampling and other generative benefits. The authors report improvements in perplexity and negative log-likelihood for next-token prediction tasks in both speech and music domains compared to purely token-based baselines. The work, spanning 6 pages with 3 figures, has been accepted to the 50th International Conference on Acoustics, Speech and Signal Processing (ICASSP) in Hyderabad, India. The preprint was first submitted in December 2024 and updated in June 2026.
What's missing
The paper reports improved perplexity and negative log-likelihood scores but does not appear to include subjective listening evaluations (e.g., MOS scores) or comparisons against the full range of state-of-the-art generative audio systems such as AudioLM or MusicGen. Computational cost and scalability of the hybrid architecture relative to discrete-only baselines are not discussed in the abstract. It is also unclear whether the improvements generalize to downstream tasks such as text-to-speech or music generation quality as perceived by human listeners.
What different sources said
- arXiv cs.LGCenter
Whisper-GPT -- Continuous Discrete Hybrid Representation Language Models For Speech And Music
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.