Dandavats,
Vrajasundari is going to do transcriptions of Guru Maharaja's classes and is interested in trying out voice recognition software. I remember that there was a discussion on Tattva-viveka about voice recognition software, but I can't find the thread. Does anyone have voice recognition software for Mac or experience on how viable it is for transcribing lectures?
Ys
Vrindaranya
Dandavats,Vrajasundari is going to do transcriptions of Guru Maharaja's classes and is interested in trying out voice recognition software. I remember that there was a discussion on Tattva-viveka about voice recognition software, but I can't find the thread. Does anyone have voice recognition software for Mac or experience on how viable it is for transcribing lectures?
Ys
Vrindaranya
Dear Vrindaranya,
That interesting topic was once covered by David Pogue, in his NY Times subscription column. I believe the PC version of the software he uses is superior to anything found on Mac. Well, if that's coming from David Pogue, a Mac guru, it means a pretty big deal.
I'll investigate more, but he himself uses that software on a Windows PC to write all his books, columns, etc. He's very happy with it.
However, considering that all Intel Macs can run Windows in Boot Camp natively (with no emulation), and also faster that Windows machines (pretty amazing, huh?), there's no reason why not transform an Intel-based Mac into a Windows dictating / transcribing machine when needed.
David says:
... As I wrote in my Vista review for The Times, I don’t find it quite as accurate (he means speech recognition built in Windows Vista) as my beloved Dragon NaturallySpeaking 9, which is freakishly, “Star Trek”-ishly accurate. But it’s awfully cool. Hook up a microphone headset and give it a shot.
So here it is, Dragon Naturally Speaking 9 is the way to go. :Bring It On:
But, there'll be lots of work, because Swami's lesson's include lots of Sanskrit terms, with complex transliteration. There's more than one language involved there. So, that transcribing effort will be a combination of both recognition and manual re-typing / editing. But very exciting work, nonetheless!
Does anyone have this program? It would be great if someone could try it out on one of the lectures to see how it does.
Does anyone have this program? It would be great if someone could try it out on one of the lectures to see how it does.
HOWEVER, there's something available for Mac, MacSpeech Dictate, and it seems that new piece of software won some award during the Macworld 2008. More about it...
Authors say:
At the core of MacSpeech Dictate is the world-renowned Dragon speech recognition engine by Nuance. It’s the brains and brawn behind MacSpeech Dictate’s phenomenal accuracy capabilities.
Here we go, a brand new thing for Mac. That sounds wonderful.
PS. (Reviews say it can recognise British English too) :Bring It On:
I tried Dragon Naturally Speaking on Guru Maharaja’s lectures a year or so ago, I think I even sent a copy to Karnam to try. The results were horrible. Several reasons might have attributed to the failure:
1. You have to train the software to recognize your voice, without it recognition factor dramatically drops. It means that Guru Maharaja need to spend a lot of time teaching software to recognize his unique voice.
2. The distance from the microphone is very important, lectures have different volume settings during recording, and Guru Maharaja does not speak directly into the microphone all the time which means that volume is changing multiple times within same lecture. Normalization is needed, but I would not make it a default for subscription CDs – they will not sound natural.
3. Speed of your speech is very important; if you speak slow recognition will go up. Guru Maharaja does not speak slowly.
4. Guru Maharaja is using a lot of lingo which we are used to (Sanskrit, Bengali verses, terminology) which is completely outside of normal vocabulary. Recognition of such words is impossible.
Now, as we know “impossible is a word in a fool’s dictionary.” Anything is possible; it just depends on effort we must invest in this. There is no magical software which will do transcription for you.
Anything is possible; it just depends on effort we must invest in this. There is no magical software which will do transcription for you.
I wanted to say the same, but didn't want to discourage anyone.
We're very specific users, and we'd need Star Trek's 24-century speech recognition software technology to satisfy our needs fully. :Bring It On: But, it can be tried, to see in which extent it can be used now.
I like the idea, but am pretty confident that transcribing the lectures by hand goes still much faster. Checking and double checking all the sanskrit verses and words, which the transcription software won't recognize will take a lot of time.
Thanks for all the responses. Although I wanted to give voice recognition software a fair chance, my intuition that it would be more hassle than help were confirmed by the feedback.