8 papers
The WER Trap: Shattering the Illusion of Unified Tokens in Speech Language Models
Xiangyu Zhang, Yuxin Li, Haoyang Zhang +5
The pursuit of a "unified" discrete token for both speech understanding and generation has led the Speech Language Model (SLM) community to heavily rely on Word Error Rate (WER) --…
Why Your Tokenizer Fails in Information Fusion: A Timing-Aware Pre-Quantization Fusion for Video-Enhanced Audio Tokenization
Xiangyu Zhang, Benjamin John Southwell, Siqi Pan +3
Audio tokenization has emerged as a critical component in end-to-end audio language models, enabling efficient discrete representation learning for both audio understanding and gen…
Distinctive Feature Codec: An Adaptive Efficient Speech Representation for Depression Detection
Xiangyu Zhang, Fuming Fang, Peng Gao +3
Large Language Models (LLMs) have demonstrated remarkable success across diverse fields, establishing a powerful paradigm for complex information processing. This has inspired the…
Zero-Shot Cognitive Impairment Detection from Speech Using AudioLLM
Mostafa Shahin, Beena Ahmed, Julien Epps
Cognitive impairment (CI) is of growing public health concern, and early detection is vital for effective intervention. Speech has gained attention as a non-invasive and easily col…
SpeechT-RAG: Reliable Depression Detection in LLMs with Retrieval-Augmented Generation Using Speech Timing Information
Xiangyu Zhang, Hexin Liu, Qiquan Zhang +2
Large Language Models (LLMs) have been increasingly adopted for health-related tasks, yet their performance in depression detection remains limited when relying solely on text inpu…
Auto-Landmark: Acoustic Landmark Dataset and Open-Source Toolkit for Landmark Extraction
Xiangyu Zhang, Daijiao Liu, Tianyi Xiao +5
In the speech signal, acoustic landmarks identify times when the acoustic manifestations of the linguistically motivated distinctive features are most salient. Acoustic landmarks h…