Unveiling the Potential of LLM-Based ASR on Chinese Open-Source Datasets
This work addresses the optimization of LLM-based ASR systems for Chinese speech recognition, representing an incremental advancement in a mainstream paradigm.
The research tackled the integration of Large Language Models (LLMs) with automatic speech recognition (ASR) on Chinese datasets, achieving state-of-the-art performance on AISHELL-1, Test_Net, and Test_Meeting test sets through a three-stage training approach and strategic component integration.
Large Language Models (LLMs) have demonstrated unparalleled effectiveness in various NLP tasks, and integrating LLMs with automatic speech recognition (ASR) is becoming a mainstream paradigm. Building upon this momentum, our research delves into an in-depth examination of this paradigm on a large open-source Chinese dataset. Specifically, our research aims to evaluate the impact of various configurations of speech encoders, LLMs, and projector modules in the context of the speech foundation encoder-LLM ASR paradigm. Furthermore, we introduce a three-stage training approach, expressly developed to enhance the model's ability to align auditory and textual information. The implementation of this approach, alongside the strategic integration of ASR components, enabled us to achieve the SOTA performance on the AISHELL-1, Test_Net, and Test_Meeting test sets. Our analysis presents an empirical foundation for future research in LLM-based ASR systems and offers insights into optimizing performance using Chinese datasets. We will publicly release all scripts used for data preparation, training, inference, and scoring, as well as pre-trained models and training logs to promote reproducible research.