Seems like they are getting close. But some issues: - Not yet Human level accuracy even for the clean speech. - Vocabulary limit is still an issue (but they a

Seems like they are getting close. But some issues: - Not yet Human level accuracy even for the clean speech. - Vocabulary limit is still an issue (but they are much better than before) - Recognition under very noisy environments. - Recognition in the existence of multiple overlapping voices - Heavy accents. - Spontaneous speech. - Languages other than English are also not in the same level.