Abstract | ||
---|---|---|
Pre-trained acoustic representations such as wav2vec and DeCoAR have attained impressive word error rates (WER) for speech recognition benchmarks, particularly when labeled data is limited. But little is known about what phonetic properties these various representations acquire, and how well they encode transferable features of speech. We compare features from two conventional and four pre-trained... |
Year | DOI | Venue |
---|---|---|
2021 | 10.1109/ICASSP39728.2021.9414776 | ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) |
Keywords | DocType | ISBN |
Voice activity detection,Semantics,Speech recognition,Phonetics,Benchmark testing,Syntactics,Signal processing | Conference | 978-1-7281-7605-5 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Danni Ma | 1 | 0 | 0.34 |
Neville Ryant | 2 | 140 | 11.00 |
Mark Liberman | 3 | 4 | 1.83 |