New AI Framework Combines Muscle Signals and Lip Reading for Silent Speech Synthesis
Researchers have introduced Listen-Write-Speak (LWS), a new paradigm enabling speech-based large language models to simultaneously listen to user audio, produce visible written text, and generate spoken responses in real time. Current speech AI systems are largely limited to verbal outputs, which prevents them from performing text-native tasks like code generation or structured analysis during live interaction. LWS aims to make visible text a first-class output channel alongside speech, potentially broadening what conversational AI can accomplish without sacrificing responsiveness.
A team of researchers has published a preprint on arXiv proposing Listen-Write-Speak (LWS), a tri-channel framework designed to overcome a fundamental limitation of speech-based large language models: their confinement to spoken outputs. In the LWS paradigm, a single autoregressive LLM concurrently listens to incoming user audio, writes free-form visible text as its primary output, and generates a real-time spoken response, all within a shared causal attention context. Crucially, this behavior is achieved entirely through a 'Token Schema' approach, requiring no changes to the underlying model architecture, and is trained via a two-stage data pipeline that synthesizes per-second cognitive annotations aligned with the input timeline. In empirical evaluations, LWS achieved a score of 4.72 on VoiceBench AlpacaEval, demonstrated strong full-duplex interaction on Full-Duplex-Bench, and reached 92.6% consistency between its written and spoken outputs, outperforming internal ablations on URO-Bench. The authors argue these results show that visible writing can function as a genuine, parallel output channel in speech interaction without degrading real-time performance, and have made their code and dataset publicly available.
What's missing
As a preprint, this work has not yet undergone formal peer review. Latency and computational cost of running three simultaneous channels in real-world deployment conditions are not discussed. The study's evaluation benchmarks (Full-Duplex-Bench, VoiceBench AlpacaEval, URO-Bench) are relatively new and their coverage of real-world use cases remains an open question.
What different sources said
- arXiv cs.AICenter
Liberating LLM Capabilities in Full-Duplex Speech Models
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.