Zero-shot Domain-sensitive Speech Recognition with Prompt-conditioning Fine-tuning

TL;DR

This paper introduces a prompt-conditioned fine-tuning approach for speech recognition that enhances domain sensitivity, achieving significant WER reductions across multiple unseen domains using both audio and text-only data.

Contribution

It presents a novel prompt-conditioning fine-tuning method for domain-sensitive speech recognition, including a text-only adaptation technique for improved domain generalization.

Findings

Up to 33% WER reduction on unseen datasets.

Effective domain adaptation with text-only fine-tuning.

Model generalizes across diverse domains and prompt contexts.

Abstract

In this work, we propose a method to create domain-sensitive speech recognition models that utilize textual domain information by conditioning its generation on a given text prompt. This is accomplished by fine-tuning a pre-trained, end-to-end model (Whisper) to learn from demonstrations with prompt examples. We show that this ability can be generalized to different domains and even various prompt contexts, with our model gaining a Word Error Rate (WER) reduction of up to 33% on unseen datasets from various domains, such as medical conversation, air traffic control communication, and financial meetings. Considering the limited availability of audio-transcript pair data, we further extend our method to text-only fine-tuning to achieve domain sensitivity as well as domain adaptation. We demonstrate that our text-only fine-tuned model can also attend to various prompt contexts, with the model reaching the most WER reduction of 29% on the medical conversation dataset.