Actually, I am just playing around with the new speech recognition function in Windows 7. I am not even touching the keyboard. Of course, the typing is not the perfect all the time. The computer makes mistakes. But the funny thing is, that if I go and correct the mistakes it is making less and less of them. And, of course, I had to train the computer a little. But it only took about 5 minutes, significantly less than anytime before. And the correction functions are much easier to use.
Because this is not the first time that I tried to use voice recognition. I played around with IBM VoiceType and ViaVoice, Dragon NaturallySpeaking, both in English and Japanese. But they were always difficult to use and not really good for any practical application.
What is interesting, that recognition works pretty well in spite of the fact that I’m not a native English speaker. I could not help registering two users of myself: an American English speaking version and a British English speaking version. Of course, I am neither: my English is somewhere in between, probably a bit more American. But I do know that I pronounce a number of words the wrong, i.e. the British way. Anyway, after the training the computer got used to my way of pronouncing words. And now it is doing surprisingly well.
Homework for tomorrow: I should try to use voice recognition in Japanese as well.
Hát sok szerencsét!
Bár csak lenne magyarul értő, beszélő verzió is!
Igaz vannak kezdeményezések, de nem túl nagy sikerrel!
A jó eredmény a jó algoritmus vagy a Moore-törvény következménye-e?
Persze, mindkettőé. De melyiké inkább?
(Mindig is imádtad a beszédfelismerőket…)
Nekem 98-99 körül volt alkalmam az OS/2-be épített beszédfelismerőt próbálni. Telepítettem, tanítottam, működött is, de aztán valahogy sose lendültem bele.
A beszédfelismerést én majd akkor hiszem el, ha egy újonnan telepített linuxos gép is képes lesz rá “out of the box”. Na akkor beérett a technológia.