10 Nov – 16 Nov 2026

This week’s shortlist

  1. Any Channel Count, One Token Stream: Incorporating Spatial Audio Understanding and Reasoning into Large Audio-Language Models
    Yuhuan You, Wenxiang Guo, Zhifang Guo, Yuxuan Wang ... · Alibaba Group +1 · arXiv
  2. CoTTAGE: An Explainable Judge for Text-to-Audio Generation
    Yu-Xiang Lin, Yun-Man Hsu, Bo-Ru Lu, Hung-yi Lee ... · National Taiwan University +2 · arXiv
  3. Emilia2: A Large-Scale Dataset for Controllable Long-Form Speech Generation
    Qinke Ni, Dekun Chen, Junwen Qiu ... · The Chinese University of Hong Kong, Shenzhen +3 · arXiv
Monday, November 16, 2026
Jin Wang, Liang Lin, Kaiwen Luo ... · Ant International +5 · arXiv

Recent advances in large audio-language models (LALMs) have broadened audio understanding across speech, sound, and music, with tool-based agents further supporting these tasks by invoking predefined audio tools. However, a persistent gap remains between perceiving acoustic…

Sunday, November 15, 2026
Jingbin Hu, Xinfa Zhu, Hao Wang, Lei Xie ... · Alibaba Group +1 · arXiv

Zero-shot voice cloning typically emphasizes preserving the timbre and prosodic characteristics of a reference speaker. However, in real-world use, speech should adapt its delivery to the target text rather than remain tied to a fixed speaking style. In this work, we formulate…

Friday, November 13, 2026
Yuhuan You, Wenxiang Guo, Zhifang Guo, Yuxuan Wang ... · Alibaba Group +1 · arXiv

Audio phase relations encode source locations, yet existing large audio-language models (LALM) reduce audio to single-channel content to stabilise general audio understanding, leaving spatial information for direction, distance, and source-relation questions unused. We present…

Yu-Xiang Lin, Yun-Man Hsu, Bo-Ru Lu, Hung-yi Lee ... · National Taiwan University +2 · arXiv

Rapid progress in text-to-audio (TTA) generation demands reliable and explainable automatic evaluation. However, existing metrics fail to simultaneously deliver high correlation with human judgments and meaningful, audio-grounded explanations, forcing reliance on labor-intensive…

Benjamin Van Niekerk, Jean-Philippe Letendre, Nicol Visser, Mirco Ravanelli ... · Ubisoft La Forge +3 · NeurIPS 2026

Speech tokenizers convert audio into discrete units, enabling language models to process and generate speech. The goal is to learn compact, text-like representations while preserving enough detail to reconstruct a speaker's voice and delivery. We propose Facto: a factorized…

Wednesday, November 11, 2026
Hengyuan Cao, Biao Gong, Guoshan Lu ... · Inclusion AI · arXiv

In this report, we introduce the Reciprocal World Model (RWM) and present Duo v1, a full-duplex world model that continuously perceives audiovisual inputs while responding to users in real time. The RWM treats intelligence and physical reality as equally fundamental to what…

Zhisheng Zhang, Yifan Mi, Yixuan Zhou ... · Shenzhen International Graduate School, Tsinghua University +3 · NeurIPS 2026

Speech tokenizers serve as speech front ends, encoding input speech signals into representations for downstream models, such as automatic speech recognition (ASR) systems and large audio-language models (LALMs). In this paper, we find that speech tokenizers are fragile at the…

Ming Ma, Mingjie Chen, Yuanzhong Zheng ... · VoxNexus · arXiv

Can a speech generator also serve as an effective voice encoder? We show that a pretrained autoregressive text-to-speech (TTS) backbone can be repurposed for reference compression, improving speaker similarity in subsequent synthesis. Jointly encoding reference speech and its…

Tuesday, November 10, 2026
Qinke Ni, Dekun Chen, Junwen Qiu ... · The Chinese University of Hong Kong, Shenzhen +3 · arXiv

Generating long-form speech requires models to keep content, speaker identity, turn-taking, and expressive style consistent over time. Yet most open training corpora contain examples lasting only tens of seconds and open long-form benchmarks cover at most two speakers per…

Tairan Chen, Richard Yucheng He, Yihang Liu ... · AveraLabs · arXiv

Full-duplex spoken dialogue, in which a system listens while it speaks and can backchannel, yield the floor, and be interrupted, is usually obtained by training on hundreds of thousands to millions of hours of speech. This is costly, rewrites the language model that supplies the…

Boyan Han, Xiaoyan Wang, Zhe Sun ... · arXiv

Answering spatial questions about multi-source scenes requires a model both to identify each sound and to locate it. Existing spatial audio-language models obtain spatial information from a dedicated encoder, which separates spatial and semantic processing into distinct feature…

Alkis Koudounas, Osamu Take, Hayato Futami ... · Sony Group Corporation · arXiv

We present Walrus, a unified 8B speech-and-audio foundation model handling speech recognition (including code-switching and multi-talker speech), speech-to-text and speech-to-speech translation, text-to-speech, emotion recognition, audio captioning, and general audio…