Exploring Phoneme-Level Speech Representations for End-to-End Speech Translation
This work addresses efficiency and performance issues in speech translation for applications like real-time communication, though it is incremental as it builds on existing phoneme-based methods.
The paper tackled the problem of long, sparse sequences in end-to-end speech translation by using compressed phoneme-level speech representations instead of frame-level features, resulting in improvements of up to 5 BLEU and a 60% reduction in training time across multiple language pairs.
Previous work on end-to-end translation from speech has primarily used frame-level features as speech representations, which creates longer, sparser sequences than text. We show that a naive method to create compressed phoneme-like speech representations is far more effective and efficient for translation than traditional frame-level speech features. Specifically, we generate phoneme labels for speech frames and average consecutive frames with the same label to create shorter, higher-level source sequences for translation. We see improvements of up to 5 BLEU on both our high and low resource language pairs, with a reduction in training time of 60%. Our improvements hold across multiple data sizes and two language pairs.