ESPnet-SpeechLM: An Open Speech Language Model Toolkit

ESPnet-SpeechLM: An Open Speech Language Model Toolkit

Stať ve sborníku v databázi WoS či Scopus

We present ESPnet-SpeechLM, an open toolkit designed to democratize the development of speech language models (SpeechLMs) and voice-driven agentic applications. The toolkit standardizes speech processing tasks by framing them as universal sequential modeling problems, encompassing a cohesive workflow of data preprocessing, pre-training, inference, and task evaluation. With ESPnet-SpeechLM, users can easily define task templates and configure key settings, enabling seamless and streamlined SpeechLM development. The toolkit ensures flexibility, efficiency, and scalability by offering highly configurable modules for every stage of the workflow. To illustrate its capabilities, we provide multiple use cases demonstrating how competitive SpeechLMs can be constructed with ESPnet-SpeechLM, including a 1.7B-parameter model pre-trained on both text and speech tasks, across diverse benchmarks.

We present ESPnet-SpeechLM, an open toolkit designed to democratize the development of speech language models (SpeechLMs) and voice-driven agentic applications. The toolkit standardizes speech processing tasks by framing them as universal sequential modeling problems, encompassing a cohesive workflow of data preprocessing, pre-training, inference, and task evaluation. With ESPnet-SpeechLM, users can easily define task templates and configure key settings, enabling seamless and streamlined SpeechLM development. The toolkit ensures flexibility, efficiency, and scalability by offering highly configurable modules for every stage of the workflow. To illustrate its capabilities, we provide multiple use cases demonstrating how competitive SpeechLMs can be constructed with ESPnet-SpeechLM, including a 1.7B-parameter model pre-trained on both text and speech tasks, across diverse benchmarks.

Speech communication; Speech processing; Data preprocessing; Language model; Model problems; Multiple use-cases; Parameter model; Pre-training; Sequential modeling; Work-flows

Speech communication; Speech processing; Data preprocessing; Language model; Model problems; Multiple use-cases; Parameter model; Pre-training; Sequential modeling; Work-flows

TIAN, J.; SHI, J.; CHEN, W.; ARORA, S.; MASUYAMA, Y.; MAEKAKU, T.; WU, Y.; PENG, J.; BHARADWAJ, S.; ZHAO, Y.; CORNELL, S.; PENG, Y.; YUE, X.; YANG, C.; NEUBIG, G.; WATANABE, S.

2026

29.04.2025

Association for Computational Linguistics (ACL)

Hybrid, Albuquerque, New Mexico, USA

9798891761919

Proceedings of the 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies: Long Papers, NAACL-HLT 2025

116

124

9

https://aclanthology.org/2025.naacl-demo.12.pdf

tian_2025.naacl-demo.12_co-author_Junyi Peng