WAVPROMPT: Towards Few-Shot Spoken Language Understanding with Frozen Language Models

TL;DR

WavPrompt leverages frozen language models and fine-tuned wav2vec to enable few-shot speech understanding, surpassing naive text baselines and extracting richer information beyond transcriptions.

Contribution

The paper introduces WavPrompt, a novel framework that adapts frozen language models for few-shot speech understanding using a fine-tuned wav2vec encoder.

Findings

WavPrompt outperforms naive text baselines in speech understanding tasks.

Detailed ablation studies identify optimal model configurations.

WavPrompt can extract additional non-speech information beyond transcriptions.

Abstract

Large-scale auto-regressive language models pretrained on massive text have demonstrated their impressive ability to perform new natural language tasks with only a few text examples, without the need for fine-tuning. Recent studies further show that such a few-shot learning ability can be extended to the text-image setting by training an encoder to encode the images into embeddings functioning like the text embeddings of the language model. Interested in exploring the possibility of transferring the few-shot learning ability to the audio-text setting, we propose a novel speech understanding framework, WavPrompt, where we finetune a wav2vec model to generate a sequence of audio embeddings understood by the language model. We show that WavPrompt is a few-shot learner that can perform speech understanding tasks better than a naive text baseline. We conduct detailed ablation studies on different components and hyperparameters to empirically identify the best model configuration. In addition, we conduct a non-speech understanding experiment to show WavPrompt can extract more information than just the transcriptions. Code is available at https://github.com/Hertin/WavPrompt