They've been saying this for... 20 years or so? Starting from the first PC soundcards? Unless AI really takes leaps I'm sure this will not happen any time soon

They've been saying this for... 20 years or so? Starting from the first PC soundcards? Unless AI really takes leaps I'm sure this will not happen any time soon. Voice is too ambiguous to convert to text with sound only. The machine will really have to understand you. Current voice recognition is really awkward and frustrating, and it's currently much easier to learn fast typing than fast voice input. (and in addition to that there are the many practical drawbacks to voice recognition as voice elsewhere in this thread)