Interactive ASR: Towards Human-Like Interaction and Semantic Coherence Evaluation for Agentic Speech Recognition
This work addresses the problem of semantic evaluation and human-like interaction in ASR for researchers and practitioners, representing a novel integration of agentic frameworks rather than an incremental improvement.
The paper tackles the limitations of Word Error Rate (WER) in evaluating semantic correctness and the lack of interactive correction in automatic speech recognition (ASR) by proposing an agentic framework that uses LLM-as-a-Judge for semantic-aware evaluation and LLM-driven multi-turn interaction for iterative refinement, achieving improved semantic fidelity and interactive correction capability on standard benchmarks.
Recent years have witnessed remarkable progress in automatic speech recognition (ASR), driven by advances in model architectures and large-scale training data. However, two important aspects remain underexplored. First, Word Error Rate (WER), the dominant evaluation metric for decades, treats all words equally and often fails to reflect the semantic correctness of an utterance at the sentence level. Second, interactive correction-an essential component of human communication-has rarely been systematically studied in ASR research. In this paper, we integrate these two perspectives under an agentic framework for interactive ASR. We propose leveraging LLM-as-a-Judge as a semantic-aware evaluation metric to assess recognition quality beyond token-level accuracy. Furthermore, we design an LLM-driven agent framework to simulate human-like multi-turn interaction, enabling iterative refinement of recognition outputs through semantic feedback. Extensive experiments are conducted on standard benchmarks, including GigaSpeech (English), WenetSpeech (Chinese), the ASRU 2019 code-switching test set. Both objective and subjective evaluations demonstrate the effectiveness of the proposed framework in improving semantic fidelity and interactive correction capability. We will release the code to facilitate future research in interactive and agentic ASR.