arXiv:2511. 20973v2 Announce Type: replace-cross Abstract: Large Audio Language Models (LALMs) deliver strong performance across speech and audio tasks, but their audio encoders generate high-rate token sequences (e.
AudioTETRIS is a training‑free, two‑stage token elimination framework designed for large audio language models. It leverages the geometry of audio encoder representations as an acoustic prior to compute per‑token importance scores, which guide progressive top‑k token selection across decoder layers. Experiments on Qwen2‑Audio‑7B, Phi‑4‑Multimodal, and Audio Flamingo 3 show that AudioTETRIS retains near‑full performance while using only 25% of audio tokens and outperforms existing training‑free baselines.
The paper introduces a compression framework for the Whisper automatic speech recognition model that jointly optimizes six deployment dimensions—model size, temporal resolution, encoder token stride, low‑rank adaptation capacity, weight precision, and sparsity pattern—using NSGA‑III. The optimization targets three objectives: word error rate, inference FLOPs, and memory footprint. Evaluating 1,680 configurations, the study identifies compression combinations that outperform single‑axis scaling and notes that 1:4 structured sparsity cannot maintain acceptable accuracy within the tested budgets.
arXiv:2607. 20086v1 Announce Type: cross Abstract: State-space sequence models are attractive for streaming speech because they maintain compact recurrent state, but scan-style training kernels can have unfavorable constants for short audio tasks.
The paper presents TTS‑Q, a post‑training quantization method that reduces the size of large diffusion transformer TTS models to 4‑bit precision while preserving speaker timbre and intelligibility. By identifying hard‑to‑quantize weights and activations and applying a mixed‑precision schedule that gives higher precision to late‑step feed‑forward activations, TTS‑Q restores timbre loss observed in standard 4‑bit PTQ. The approach also offers a calibrated‑W4A4 reference that lowers GPU memory usage, making TTS diffusion transformers more viable for on‑device, assistive, and real‑time applications.
arXiv:2608. 08569v1 Announce Type: new Abstract: Recent advancements in Speech Large Language Models have demonstrated remarkable capabilities in understanding complex audio tasks.
The paper introduces TTS‑Q, a perception‑guided post‑training quantization method for zero‑shot text‑to‑speech diffusion transformers. It identifies that 4‑bit weight‑and‑activation quantization erodes speaker timbre, and addresses this by selecting hard‑to‑quantize items and applying a mixed‑precision schedule that gives higher precision to late‑step feed‑forward activations. The resulting approach restores timbre while preserving intelligibility and reduces GPU memory usage for resource‑constrained deployment.
arXiv:2509.10452v3 Announce Type: replace-cross Abstract: Pretrained automatic speech recognition (ASR) models such as Whisper perform well but still need domain adaptation to handle unseen parlance....
The paper introduces asymmetric classifier‑free guidance (CFG) for target‑speaker ASR using Whisper, where a speaker‑conditioned branch predicts the target transcript and a speaker‑unconditioned branch predicts serialized multi‑speaker transcripts. CFG modulates the influence of speaker conditioning during decoding via a single guidance scale, which is first set globally on development data and then refined per utterance by a lightweight encoder‑based predictor while keeping the recognition model fixed. The resulting system yields up to 21.8% relative WER reduction over a condition‑only baseline and 5.6% over standard conditional decoding under domain shifts.
The paper introduces AWED, a word‑level emission‑delay metric, and demonstrates that post‑training a delayed‑streaming model with a joint reward (GRPO) improves both accuracy and latency. Using a single operating point (τ=6 frames), the method outperforms both its supervised baseline and the Voxtral Realtime backbone across all lookahead budgets, reducing WER by up to 30.8% at 80 ms delay and lowering median AWED from 1.17 s to 1.04 s.
arXiv:2606. 01483v1 Announce Type: cross Abstract: Long-form automatic speech recognition (ASR) requires both high accuracy and low latency, but existing systems force a trade-off between the two.
Neural audio codecs are a key component of speech processing pipelines, compressing audio into discrete tokens for downstream modeling. However, existing codecs struggle to balance reconstruction quality with token efficiency, often encoding perceptually irrelevant information such as background noise and recording artifacts at the expense of linguistically and acoustically meaningful content.