Self-supervised learning (SSL)-based speech models are extensively used for full-stack speech processing. However, it has been observed that improving SSL-based speech representations using unlabeled speech for content-related tasks is challenging and computationally expensive. Recent attempts have been made to address this issue with cost-effective self-supervised fine-tuning (SSFT) approaches. Continuing in this direction, a cost-effective SSFT method named "LASER: Learning by Aligning Self-supervised Representations" is presented. LASER is based on the soft-DTW alignment loss with temporal regularisation term. Experiments are conducted with HuBERT and WavLM models and evaluated on the SUPERB benchmark for two content-related tasks: automatic speech recognition (ASR) and phoneme recognition (PR). A relative improvement of 3.7% and 8.2% for HuBERT, and 4.1% and 11.7% for WavLM are observed, for the ASR and PR tasks respectively, with only < 3 hours of fine-tuning on a single GPU.
Refining Self-Supervised Learnt Speech Representation using Brain Activations
National Engineering Research Center of Speech and Language Information Processing（国家语言信息处理工程研究中心）；University of Science and Technology of China（中国科学技术大学）
VATLM: Visual-Audio-Text Pre-Training with Unified Masked Prediction for Speech Representation Learning
National Engineering Research Center of Speech and Language Information Processing（国家语音及语言信息处理工程技术研究中心）；University of Science and Technology of China（中国科学技术大学）；Microsoft Corporation（微软公司）
Zipper-LoRA: Dynamic Parameter Decoupling for Speech-LLM based Multilingual Speech Recognition
Shanghai Engineering Research Center of Intelligent Education and Bigdata, Shanghai Normal University（智能教育与大数据上海工程研究中心，上海师范大学）；SHNU-Unisound Natural Human-Computer Interaction Lab, Shanghai Normal University（上海师范大学Unisound自然人机交互实验室）；Unisound AI Technology Co., Ltd.（Unisound人工智能技术有限公司）
Adaptive Inner Speech-Text Alignment for LLM-based Speech Translation
Institute of Computing and Intelligence, Harbin Institute of Technology（哈尔滨工业大学计算与智能学院）；School of Computer Science and Engineering, Xi’an University of Technology（西安理工大学计算机科学与工程学院）
Revealing the Role of Audio Channels in ASR Performance Degradation
Dept. Computer Science and Information Engineering, National Taiwan Normal University, Taiwan（计算机科学与信息工程系，台湾国立台湾师范大学）；Dept. Computer Science and Information Engineering, National Tsing Hua University, Taiwan（计算机科学与信息工程系，台湾国立清华大学）；Institute of Computer Science, Academia Sinica, Taiwan（学术院计算机科学研究所，台湾）；United Link Co., Ltd., Taiwan（联合链接公司，台湾）
本文提出一种特征归一化技术，通过将 ASR 内部表示与干净参考通道对齐，缓解录音通道变化导致的性能退化，并在未见通道和语言上显著提升 ASR 性能。
Detecting Alzheimer's Disease by estimating attention and elicitation path through the alignment of spoken picture descriptions with the picture prompt
University of Sheffield（谢菲尔德大学）；Royal Hallamshire Hospital（皇家哈勒姆医院）；Sheffield Institute for Translational Neuroscience（谢菲尔德转化神经科学研究所）
HPRO: Hierarchical Progressive Reward Optimization via Preference Extraction for Emotional Text-to-Speech
South China University of Technology, China（华南理工大学，中国）；Huya Inc., China（欢丫公司，中国）；Tongyi Fun Team, Alibaba Group, China（通义 Fun 团队，阿里巴巴集团，中国）；Foshan University, China（佛山大学，中国）
FuseCodec: Semantic-Contextual Fusion and Supervision for Neural Codecs
Center for Computational & Data Sciences, Independent University, Bangladesh（计算与数据科学中心，独立大学，孟加拉国）；Amazon GenAI（亚马逊生成人工智能）；Qatar Computing Research Institute（卡塔尔计算研究所）；University of Virginia（弗吉尼亚大学）
