arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2410.03751cs.CLcs.SDeess.AS
收藏

语音语言模型的最新进展:综述

Recent Advances in Speech Language Models: A Survey

Department of Computer Science and Engineering, The Chinese University of Hong Kong(计算机科学与工程系,香港中文大学);LIGHTSPEED STUDIOS;National University of Singapore(新加坡国立大学);Tencent(腾讯)

Wenqian Cui, Dianzhi Yu, Xiaoqi Jiao, Ziqiao Meng, Guangyan Zhang, Qichao Wang, Yiwen Guo, Irwin King

AI总结:

本综述首次全面概述语音语言模型(SpeechLMs)的构建方法、架构组件、训练方案、能力、评估指标及未来挑战,旨在替代ASR+LLM+TTS流水线,实现端到端语音生成。

摘要:

大型语言模型(LLMs)近期引起了广泛关注,主要因其在基于文本的交互中的能力。然而,自然的人类交互往往依赖于语音,这促使向基于语音的模型转变。实现这一目标的一种直接方法涉及“自动语音识别(ASR)+ LLM + 文本到语音(TTS)”的流水线,其中输入语音被转录为文本,由LLM处理,然后转换回语音。尽管这种方法直接,但它存在固有的局限性,例如模态转换期间的信息丢失、由于复杂流水线导致的显著延迟,以及三个阶段中的错误累积。为了解决这些问题,语音语言模型(SpeechLMs)——无需从文本转换即可生成语音的端到端模型——已成为一种有前景的替代方案。本综述论文首次全面概述了构建SpeechLMs的最新方法,详细介绍了其架构的关键组成部分以及对其开发至关重要的各种训练方案。此外,我们系统地调查了SpeechLMs的各种能力,对其评估指标进行了分类,并讨论了这一快速发展的领域中的挑战和未来研究方向。GitHub仓库可在https://github.com/dreamtheater123/Awesome-SpeechLM-Survey获取。

英文摘要:

Large Language Models (LLMs) have recently garnered significant attention, primarily for their capabilities in text-based interactions. However, natural human interaction often relies on speech, necessitating a shift towards voice-based models. A straightforward approach to achieve this involves a pipeline of ``Automatic Speech Recognition (ASR) + LLM + Text-to-Speech (TTS)", where input speech is transcribed to text, processed by an LLM, and then converted back to speech. Despite being straightforward, this method suffers from inherent limitations, such as information loss during modality conversion, significant latency due to the complex pipeline, and error accumulation across the three stages. To address these issues, Speech Language Models (SpeechLMs) -- end-to-end models that generate speech without converting from text -- have emerged as a promising alternative. This survey paper provides the first comprehensive overview of recent methodologies for constructing SpeechLMs, detailing the key components of their architecture and the various training recipes integral to their development. Additionally, we systematically survey the various capabilities of SpeechLMs, categorize their evaluation metrics, and discuss the challenges and future research directions in this rapidly evolving field. The GitHub repository is available at https://github.com/dreamtheater123/Awesome-SpeechLM-Survey

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑