On the Transferability of Whisper-based Representations for "In-the-Wild" Cross-Task Downstream Speech Applications
This work addresses the problem of cross-task and real-world deployment of speech models for researchers and practitioners, but it is incremental as it applies an existing method to new data and conditions.
The paper investigated the transferability of Whisper's representations to four other speech tasks in the SUPERB benchmark and their robustness in noisy, real-world conditions, finding that Whisper achieves promising results across tasks and environments.
Large self-supervised pre-trained speech models have achieved remarkable success across various speech-processing tasks. The self-supervised training of these models leads to universal speech representations that can be used for different downstream tasks, ranging from automatic speech recognition (ASR) to speaker identification. Recently, Whisper, a transformer-based model was proposed and trained on large amount of weakly supervised data for ASR; it outperformed several state-of-the-art self-supervised models. Given the superiority of Whisper for ASR, in this paper we explore the transferability of the representation for four other speech tasks in SUPERB benchmark. Moreover, we explore the robustness of Whisper representation for ``in the wild'' tasks where speech is corrupted by environment noise and room reverberation. Experimental results show Whisper achieves promising results across tasks and environmental conditions, thus showing potential for cross-task real-world deployment.