ImagineNET: Target Speaker Extraction with Intermittent Visual Cue through Embedding Inpainting

TL;DR

This paper introduces ImagineNET, a novel speaker extraction method that effectively handles intermittent visual cues by jointly performing visual embedding inpainting and speaker extraction, improving robustness and performance in real-world scenarios.

Contribution

It proposes a joint framework with an interlacing structure for speaker extraction and visual inpainting, addressing visual occlusion issues in audio-visual speaker separation.

Findings

Outperforms baseline in signal quality

Enhances perceptual quality and intelligibility

Effective with different visual embeddings

Abstract

The speaker extraction technique seeks to single out the voice of a target speaker from the interfering voices in a speech mixture. Typically an auxiliary reference of the target speaker is used to form voluntary attention. Either a pre-recorded utterance or a synchronized lip movement in a video clip can serve as the auxiliary reference. The use of visual cue is not only feasible, but also effective due to its noise robustness, and becoming popular. However, it is difficult to guarantee that such parallel visual cue is always available in real-world applications where visual occlusion or intermittent communication can occur. In this paper, we study the audio-visual speaker extraction algorithms with intermittent visual cue. We propose a joint speaker extraction and visual embedding inpainting framework to explore the mutual benefits. To encourage the interaction between the two tasks, they are performed alternately with an interlacing structure and optimized jointly. We also propose two types of visual inpainting losses and study our proposed method with two types of popularly used visual embeddings. The experimental results show that we outperform the baseline in terms of signal quality, perceptual quality, and intelligibility.