Relative Positional Encoding for Speech Recognition and Direct Translation
| Autor: | Tuan-Nam Nguyen, Ngoc-Quan Pham, Alex Waibel, Thai-Son Nguyen, Thanh-Le Ha, Sebastian Stüker, Jan Niehues, Elizabeth Salesky |
|---|---|
| Přispěvatelé: | Dept. of Advanced Computing Sciences, RS: FSE DACS, DKE Scientific staff |
| Rok vydání: | 2020 |
| Předmět: |
FOS: Computer and information sciences
Sound (cs.SD) Computer Science - Computation and Language Audio and Speech Processing (eess.AS) Computer science Speech recognition FOS: Electrical engineering electronic engineering information engineering Encoding (semiotics) Translation (geometry) Computation and Language (cs.CL) Computer Science - Sound Electrical Engineering and Systems Science - Audio and Speech Processing |
| Zdroj: | INTERSPEECH INTERSPEECH 2020 Proceedings, 31-35 STARTPAGE=31;ENDPAGE=35;TITLE=INTERSPEECH 2020 Proceedings |
| Popis: | Transformer models are powerful sequence-to-sequence architectures that are capable of directly mapping speech inputs to transcriptions or translations. However, the mechanism for modeling positions in this model was tailored for text modeling, and thus is less ideal for acoustic inputs. In this work, we adapt the relative position encoding scheme to the Speech Transformer, where the key addition is relative distance between input states in the self-attention network. As a result, the network can better adapt to the variable distributions present in speech data. Our experiments show that our resulting model achieves the best recognition result on the Switchboard benchmark in the non-augmentation condition, and the best published result in the MuST-C speech translation benchmark. We also show that this model is able to better utilize synthetic data than the Transformer, and adapts better to variable sentence segmentation quality for speech translation. Submitted to Interspeech 2020 |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|