Výsledky vyhledávání - "Shan, Changhao"

Report

Streaming Decoder-Only Automatic Speech Recognition with Discrete Speech Units: A Pilot Study

Autor: Chen, Peikun, Sun, Sining, Shan, Changhao, Yang, Qing, Xie, Lei

Unified speech-text models like SpeechGPT, VioLA, and AudioPaLM have shown impressive performance across various speech-related tasks, especially in Automatic Speech Recognition (ASR). These models typically adopt a unified method to model discrete s

Externí odkaz: http://arxiv.org/abs/2406.18862

Zobrazit plný text záznamu

Report

Skipformer: A Skip-and-Recover Strategy for Efficient Speech Recognition

Autor: Zhu, Wenjing, Sun, Sining, Shan, Changhao, Fan, Peng, Yang, Qing

Conformer-based attention models have become the de facto backbone model for Automatic Speech Recognition tasks. A blank symbol is usually introduced to align the input and output sequences for CTC or RNN-T models. Unfortunately, the long input lengt

Externí odkaz: http://arxiv.org/abs/2403.08258

Zobrazit plný text záznamu

Report

Key Frame Mechanism For Efficient Conformer Based End-to-end Speech Recognition

Autor: Fan, Peng, Shan, Changhao, Sun, Sining, Yang, Qing, Zhang, Jianwei

Recently, Conformer as a backbone network for end-to-end automatic speech recognition achieved state-of-the-art performance. The Conformer block leverages a self-attention mechanism to capture global information, along with a convolutional neural net

Externí odkaz: http://arxiv.org/abs/2310.14954

Zobrazit plný text záznamu

Report

A meta learning scheme for fast accent domain expansion in Mandarin speech recognition

Autor: Zhu, Ziwei, Shan, Changhao, Zhang, Bihong, Yu, Jian

Spoken languages show significant variation across mandarin and accent. Despite the high performance of mandarin automatic speech recognition (ASR), accent ASR is still a challenge task. In this paper, we introduce meta-learning techniques for fast a

Externí odkaz: http://arxiv.org/abs/2307.12262

Zobrazit plný text záznamu