2021 IEEE International Conference on Acoustics, Speech and Signal Processing
6-11 June 2021 • Toronto, Ontario, Canada
Extracting Knowledge from Information
| Paper ID | SPE-31.2 | ||
| Paper Title | MUTUALLY-CONSTRAINED MONOTONIC MULTIHEAD ATTENTION FOR ONLINE ASR | ||
| Authors | Jaeyun Song, Hajin Shim, Eunho Yang, Korea Advanced Institute of Science and Technology (KAIST), South Korea | ||
| Session | SPE-31: Speech Recognition 11: Novel Approaches | ||
| Location | Gather.Town | ||
| Session Time: | Thursday, 10 June, 13:00 - 13:45 | ||
| Presentation Time: | Thursday, 10 June, 13:00 - 13:45 | ||
| Presentation | Poster | ||
| Topic | Speech Processing: [SPE-GASR] General Topics in Speech Recognition | ||
| IEEE Xplore Open Preview | Click here to view in IEEE Xplore | ||
| Abstract | Despite the feature of real-time decoding, Monotonic Multihead Attention (MMA) shows comparable performance to the state-of-the-art offline methods in machine translation and automatic speech recognition (ASR) tasks. However, the latency of MMA is still a major issue in ASR and should be combined with a technique that can reduce the test latency at inference time, such as head-synchronous beam search decoding, which forces all non-activated heads to activate after a small fixed delay from the first head activation. In this paper, we remove the discrepancy between training and test phases by considering, in the training of MMA, the interactions across multiple heads that will occur in the test time. Specifically, we derive the expected alignments from monotonic attention by considering the boundaries of other heads and reflect them in the learning process. We validate our proposed method on the two standard benchmark datasets for ASR and show that our approach, MMA with the mutually-constrained heads from the training stage, provides better performance than baselines. | ||