Detection of Prosodic Boundaries in Speech Using Wav2Vec 2.0

TL;DR

This paper demonstrates that wav2vec 2.0 can effectively detect prosodic boundaries in speech using only acoustic data, outperforming text-based methods and benefiting from combined approaches.

Contribution

It introduces a novel application of wav2vec 2.0 for prosodic boundary detection in speech, showing high accuracy with limited labeled data.

Findings

Wav2vec 2.0 achieves 94% accuracy on within-sentence boundaries.

The acoustic model outperforms a text-based predictor.

Combining acoustic and text-based models improves boundary detection results.

Abstract

Prosodic boundaries in speech are of great relevance to both speech synthesis and audio annotation. In this paper, we apply the wav2vec 2.0 framework to the task of detecting these boundaries in speech signal, using only acoustic information. We test the approach on a set of recordings of Czech broadcast news, labeled by phonetic experts, and compare it to an existing text-based predictor, which uses the transcripts of the same data. Despite using a relatively small amount of labeled data, the wav2vec2 model achieves an accuracy of 94% and F1 measure of 83% on within-sentence prosodic boundaries (or 95% and 89% on all prosodic boundaries), outperforming the text-based approach. However, by combining the outputs of the two different models we can improve the results even further.