Litcius/Paper detail

Adaptive Feature Selection for End-to-End Speech Translation

Biao Zhang, Ivan Titov, Barry Haddow, Rico Sennrich

202033 citationsDOIOpen Access PDF

Abstract

Information in speech signals is not evenly distributed, making it an additional challenge for end-to-end (E2E) speech translation (ST) to learn to focus on informative features. In this paper, we propose adaptive feature selection (AFS) for encoder-decoder based E2E ST. We first pre-train an ASR encoder and apply AFS to dynamically estimate the importance of each encoded speech feature to ASR. A ST encoder, stacked on top of the ASR encoder, then receives the filtered features from the (frozen) ASR encoder. We take L 0 DROP (Zhang et al., 2020) as the backbone for AFS, and adapt it to sparsify speech features with respect to both temporal and feature dimensions. Results on LibriSpeech En-Fr and MuST-C benchmarks show that AFS facilitates learning of ST by pruning out 84% temporal features, yielding an average translation gain of 1.3-1.6 BLEU and a decoding speedup of 1.4. In particular, AFS reduces the performance gap compared to the cascade baseline, and outperforms it on LibriSpeech En-Fr with a BLEU score of 18.56 (without data augmentation). 1

Topics & Concepts

Computer scienceEncoderSpeech recognitionDecoding methodsEnd-to-end principleFeature (linguistics)PruningEncoding (memory)Focus (optics)Artificial intelligenceSelection (genetic algorithm)CascadeAlgorithmEngineeringOpticsPhilosophyBiologyPhysicsLinguisticsAgronomyChemical engineeringOperating systemNatural Language Processing TechniquesTopic ModelingSpeech Recognition and Synthesis
Adaptive Feature Selection for End-to-End Speech Translation | Litcius