There are new, very large public English speech datasets: Mozilla Common Voice, National Speech Corpus, which can be combined with LibriSpeech to train large models.
Actually end-2-end CTC models are very good - Wav2letter, Jasper, QuartzNet - all these models are much better than DeepSpeech2.
There are new, very large public English speech datasets: Mozilla Common Voice, National Speech Corpus, which can be combined with LibriSpeech to train large models.
Actually end-2-end CTC models are very good - Wav2letter, Jasper, QuartzNet - all these models are much better than DeepSpeech2.