literally flawless speech recognition with natural language parsing for commands
No way. I can type much, much faster than I can speak. Simpler, strict non-natural commands are much easier than natural language commands (hell, people have difficulties giving and understanding verbal commands all the time) - never mind the fact that languages like English are terribly ambiguous and peoples everyday speech is too filled with local slang.
Also, if I have to talk to my computer, I a) cannot use it in a group environment or, say, when someone is sleeping and b) when I type I can see what I typed and I can edit it if I make mistakes. If I speak commands, I cannot do this. An exaggerated (but still realistic) example: if I type "delete foo" instead of "delete bar" and notice before I hit enter, I can fix the mistake. If I speak "delete foo" then foo will probably be gone before I can correct myself. Finally, what about different words that sound the same? Sometimes spelling or capitalization is the only thing that distinguishes things (eg, file names) - with speech you couldn't do this.
In summary, in my opinion, speech recognition is a terrible form of input for general text input and command input tasks. (It may be useful to augment more traditional input and it can certainly be more than useful (or even required) for disabled people, but for the majority of people and tasks, I don't see the attraction)
It's not that I like speech recognition, it's just the only other option that is even close to a keyboard that I can think of besides a direct brain to machine link.
No way. I can type much, much faster than I can speak. Simpler, strict non-natural commands are much easier than natural language commands (hell, people have difficulties giving and understanding verbal commands all the time) - never mind the fact that languages like English are terribly ambiguous and peoples everyday speech is too filled with local slang.
Also, if I have to talk to my computer, I a) cannot use it in a group environment or, say, when someone is sleeping and b) when I type I can see what I typed and I can edit it if I make mistakes. If I speak commands, I cannot do this. An exaggerated (but still realistic) example: if I type "delete foo" instead of "delete bar" and notice before I hit enter, I can fix the mistake. If I speak "delete foo" then foo will probably be gone before I can correct myself. Finally, what about different words that sound the same? Sometimes spelling or capitalization is the only thing that distinguishes things (eg, file names) - with speech you couldn't do this.
In summary, in my opinion, speech recognition is a terrible form of input for general text input and command input tasks. (It may be useful to augment more traditional input and it can certainly be more than useful (or even required) for disabled people, but for the majority of people and tasks, I don't see the attraction)