3 ms·
If I may insert a relevant plug: we (MERL) just put out a paper last week with SOTA 7.0 % WER on LibriSpeech test-other (vs wav2letter@anywhere's 7.5%) with 590
by jonathanleroux 7y ago
If I may insert a relevant plug: we (MERL) just put out a paper last week with SOTA 7.0 % WER on LibriSpeech test-other (vs wav2letter@anywhere's 7.5%) with 590 ms theoretical latency using joint CTC-Transformer with parallel time-delayed LSTM and triggered attention.
Check it out: https://arxiv.org/abs/2001.02674 https://arxiv.org/abs/2001.02674
- jonathanleroux 7y agoCorrection: no PTDLSTM here but time-restricted self-attention (duh). PTDLSTM was our previous encoder setup published at ASRU.