A Glance is Enough: Extract Target Sentence By Looking at A keyword
This addresses speech extraction for applications like social security, though it appears incremental as it adapts existing Transformer mechanisms to a specific task.
The paper tackles the problem of extracting target sentences from multi-talker speech using only a keyword as input, such as identifying a speaker's utterance containing 'help' in noisy environments. The proposed Transformer-based method achieves a phone error rate of 26% at SNR=-3dB, significantly outperforming a baseline with 96% PER.
This paper investigates the possibility of extracting a target sentence from multi-talker speech using only a keyword as input. For example, in social security applications, the keyword might be "help", and the goal is to identify what the person who called for help is articulating while ignoring other speakers. To address this problem, we propose using the Transformer architecture to embed both the keyword and the speech utterance and then rely on the cross-attention mechanism to select the correct content from the concatenated or overlapping speech. Experimental results on Librispeech demonstrate that our proposed method can effectively extract target sentences from very noisy and mixed speech (SNR=-3dB), achieving a phone error rate (PER) of 26\%, compared to the baseline system's PER of 96%.