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.
By Vyom Agarwal, Mokshda Gangrade, Siddharth Pal, Jerry Wu
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.
By Saurabhchand Bhati, Samuel Thomas, Hilde Kuehne, Rogerio Feris, James Glass
UniAE-MoE is a unified audio encoder that uses a Mixture-of-Experts architecture to model cross‑domain audio representations. It integrates encoder components from Qwen2‑Audio and Audio‑Flamingo 3, enhances them with SwiGLU and shared experts, and applies a two‑stage instruction‑tuning strategy along with task‑specific data scaling. The model achieves state‑of‑the‑art results on the XARES‑LLM benchmark (0.802) and tops the Interspeech 2026 Audio Encoder Capability Challenge, demonstrating strong generalization across speech, music, and general audio tasks.
By Shengbo Cai, Zhisheng Zhang, Zichao Nie, Jing Peng, Jingran Xie, Zhiyong Wu
arXiv:2608. 08569v1 Announce Type: new Abstract: Recent advancements in Speech Large Language Models have demonstrated remarkable capabilities in understanding complex audio tasks.
By Wenxu Jia, Dongjie Fu, Xize Cheng, Fangming Feng, Linjun Li, Wenshi Chen, Yingming Li, Zhou Zhao, Tao Jin
The paper introduces a reparameterization technique that injects feature noise to jointly optimize speech model performance and computational complexity during training. Unlike traditional pruning, this method dynamically adjusts model size for a desired performance‑complexity trade‑off without heuristic weight removal. The authors validate their approach with a synthetic example and two real‑world applications—voice activity detection and audio anti‑spoofing—providing publicly available code for further research.
By Esteban G\'omez, Tom B\"ackstr\"om
arXiv:2608. 04351v1 Announce Type: cross Abstract: Large Audio-Language Models (LALMs) excel at general speech understanding; however, adapting them to fine-grained tasks like Speech Emotion Recognition (SER) remains a significant bottleneck.
By Tian Jin, Ruikang Zhang, Zefeng Zhao, Ding Luo, Jin Zeng
Large Audio-Language Models (LALMs) excel at general speech understanding; however, adapting them to fine-grained tasks like Speech Emotion Recognition (SER) remains a significant bottleneck. Current Parameter-Efficient Fine-Tuning (PEFT) methods typically operate in flat Euclidean space, and this geometry fails to capture the multi-granularity nature of emotion cues, which range from low-level prosody to high-level semantics.
arXiv:2512.07571v3 Announce Type: replace
Abstract: This paper presents a simple method that allows to easily enhance textual pre-trained large language models with speech information, when fine-tune...
By Nicolas Calbucura, Jose Guillen, Valentin Barriere
arXiv:2608. 19936v1 Announce Type: cross Abstract: Public benchmarks are important measures of Automatic Speech Recognition (ASR) model capabilities.
By Theo Lebryk, David Ayllon, Alice Baird, Jakub Piotr C{\l}apa, Jens Madsen, Panagiotis Tzirakis
TontaubeV1 is a streaming text‑to‑speech model that preserves natural prosody while running on a single consumer GPU. It encodes speech with a hierarchical DualCodec representation at 12.5 Hz, separating a semantic stream from successive acoustic refinements, and uses Qwen3‑derived transformers to predict the semantic stream, utterance duration, and acoustic refinements. The system supports up to one minute of reference audio for voice conditioning, streams with a 200 ms latency to first audio, and achieves real‑time factors of 0.08 (single input) and 0.02 (eight concurrent inputs).
By Fritz Cremer, Jonathan Cremer
The paper presents a new evaluation protocol for post‑training quantization of speech language models that separates lexical output, transcript‑insufficient endpoints, and packed implementations. In a Qwen2‑Audio case study, a 6‑bit allocation selected for translation improves chrF scores but degrades emotion recognition, while uniform and front‑layer controls perform better on emotion tasks. Similar patterns hold at 7 bits, and a 4‑bit study shows consistent emotion deficits across all low‑bit allocations, with no advantage for the selected scheme. The study highlights a precision‑dependent mismatch between lexical output, waveform‑dependent behavior, and nominal precision, without claiming a general failure of low‑bit models or a deployment benefit for the selected allocation.
By Mengzhe Geng, Jinxi Jin, Junhao Xu
The paper introduces modality‑gated deep adapters, a parameter‑efficient method for adding new modalities to a frozen multimodal embedding language model without altering its existing outputs. These adapters are bottleneck modules attached to each decoder layer, grouped into modality‑specific packs that activate only during encoding of their own modality, ensuring exact preservation of the base model’s computation graph. Experiments on a 2B base model show significant gains in audio‑to‑text and thermal‑to‑text retrieval metrics, and the authors release the audio and thermal packs along with training and evaluation code.
By Abdul Basit Tonmoy, Kazi Fardinul Hoque, Md. Shahrier Islam Arham, Arman Luthra