2 ms·
It was used quite a bit in WFST-based speech recognition decoding, before so-called end-to-end approaches (CTC, RNN-T, AED, etc) took over.
by woodson 2y ago
It was used quite a bit in WFST-based speech recognition decoding, before so-called end-to-end approaches (CTC, RNN-T, AED, etc) took over.