On The Landscape of Spoken Language Models: A Comprehensive Survey
On The Landscape of Spoken Language Models: A Comprehensive Survey [144.1] 音声言語モデル(SLM)は、普遍的な音声処理システムとして機能する。 この領域での作業は非常に多様であり、様々な用語と評価設定がある。 論文参考訳（メタデータ） (Fri, 11 Apr 2025 13:40:53 GMT)
「In the last few years, the field of natural language processing (NLP) has evolved from (1) training many task-specific models from scratch, to (2) combining pre-trained multi-purpose contextual representation models (such as BERT (Devlin et al , 2019)) with a small number of task-specific parameters, to (3) training generative universal, large language models (LLMs (Brown et al , 2020; OpenAI et al , 2024)1) that perform arbitrary text tasks given natural language instructions (prompts) and can generalize to unseen domains and tasks (Wei et al , 2022a; Liu et al , 2023), and finally to (4) dialogue / chatbot systems that function as assistants and perform tasks while directly interacting with the user.」、「The field of speech processing has been undergoing a similar evolution, although with some lag, and has mainly focussed on stages (1) and (2).」から始まるspoken language models (SLMs) のサーベイ。