pTSE-T: Presentation Target Speaker Extraction using Unaligned Text Cues

TL;DR

This paper introduces pTSE-T, a novel target speaker extraction method that uses unaligned text cues from presentations to improve speech separation in challenging scenarios.

Contribution

It proposes a new approach that leverages semantic cues from limited, unaligned text to enhance target speaker extraction without relying on pre-recorded or visual information.

Findings

Achieved SI-SDRi of 12.16 dB in experiments.

Demonstrated effectiveness of semantic cues in noisy environments.

Outperformed baseline methods in target speaker extraction accuracy.

Abstract

Target Speaker Extraction (TSE) aims to extract the clean speech of the target speaker in an audio mixture, eliminating irrelevant background noise and speech. While prior work has explored various auxiliary cues including pre-recorded speech, visual information, and spatial information, the acquisition and selection of such strong cues are infeasible in many practical scenarios. Differently, in this paper, we condition the TSE algorithm on semantic cues extracted from limited and unaligned text contents, such as condensed points from a presentation slide. This method is particularly useful in scenarios like meetings, poster sessions, or lecture presentations, where acquiring other cues in real time may be challenging. To this end, we design two different networks. Specifically, our proposed Text Prompt Extractor Network (TPE) fuses audio features with content-based semantic cues to facilitate time-frequency mask generation to filter out extraneous noise. The experimental results show the efficacy in accurately extracting the target speaker's speech by utilizing semantic cues derived from limited and unaligned text, resulting in SI-SDRi of 12.16 dB, SDRi of 12.66 dB, PESQi of 0.830 and STOIi of 0.150.