Two Heads Are Better Than One: Audio-Visual Speech Error Correction with Dual Hypotheses
This addresses the problem of robust speech recognition in noisy environments for applications like assistive technologies, representing a novel paradigm rather than an incremental improvement.
The paper tackles error correction in audio-visual speech recognition by introducing a generative framework that combines independent hypotheses from separate ASR and VSR models using a large language model, achieving up to 57.7% error rate gain on the LRS2 benchmark over a standard ASR baseline.
This paper introduces a new paradigm for generative error correction (GER) framework in audio-visual speech recognition (AVSR) that reasons over modality-specific evidences directly in the language space. Our framework, DualHyp, empowers a large language model (LLM) to compose independent N-best hypotheses from separate automatic speech recognition (ASR) and visual speech recognition (VSR) models. To maximize the effectiveness of DualHyp, we further introduce RelPrompt, a noise-aware guidance mechanism that provides modality-grounded prompts to the LLM. RelPrompt offers the temporal reliability of each modality stream, guiding the model to dynamically switch its focus between ASR and VSR hypotheses for an accurate correction. Under various corruption scenarios, our framework attains up to 57.7% error rate gain on the LRS2 benchmark over standard ASR baseline, contrary to single-stream GER approaches that achieve only 10% gain. To facilitate research within our DualHyp framework, we release the code and the dataset comprising ASR and VSR hypotheses at https://github.com/sungnyun/dualhyp.