My job involves a lot of travel be it plane or auto. I also need to do a lot
of call reports relating to my work. I am a cumbersome typist and if I had a
way to reduce typing boy would it ever make my job easier. I was reading your
article about voice recognition a while back and it got me to thinking. I have
a battery operated mp3 player/voice recorder. Could I do dictation to the voice
recorder while traveling our in my motel room and then set up my computer so it
could do voice recognition right through the sound card in my computer using my
voice recorder as the source of the voice? If so how would I set this up?
I’ll give you an idea or two about setting it up, but then I’ll advise you
not to bother.
My belief is with the state of today’s voice recognition technologies, down
this path lies only frustration for the situation you describe.
Become a Patron of Ask Leo! and go ad-free!
Voice recognition today is still very touchy. It’s better than it
was, but still not quite to the level we might want it to be.
It really works best with a good, clean sounding source. I would think that
adding a voice recorder as an intermediate storage device seems like it would
add a bit of noise and increase the error rate to an unacceptable level. And
recording in a car or airplane is definitely going to involve some noise.
When using voice recording “live”, you can sort of watch what’s being
“heard” and make corrections on the way. It might slow you down a little, but
it can be manageable.
If you’re attempting voice recognition from a recording, you have no
options. The recording is going to plow ahead no matter what errors are
mis-recognized. My belief is you’ll spend as much time correcting the resulting
text as you would have typing it in to start.
But, for the record, here’s how I’d try it:
Get a stereo “patch cord” that you can then use to plug the headphone out on
the voice recorder to the into the line in on your computer’s sound card.
Alternately, use that same patch cord to headphone out to the computer’s
microphone in. (And use lower volume settings as we proceed below.)
Set up the voice recognition software on your PC for dictation, possibly
transcribing into something like a Word document.
Press play on your voice recorder.
In concept it’s very simple. In practice, not so much. There’s going to be a
lot of trial and error before you figure out what speaking speed works best,
as well as what recording and playback volume and other settings work best. On top of
that, most voice recognition software need to be “trained”, which means that
you’ll need to spend some time recording some stock speech that the software
would then listen to in this manner to learn what it is you sound like.
What most folks do in a situation like this is use a human transcriber.
There are still things that only a human can do effectively, and for now this
is still one of them.
Many voice recorders actually save natively now in a computer audio format.
I particularly like my Olympus
WS-100 because it records to Windows “wma” format and has a USB interface;
plug it in and it’s just another drive on my system with a bunch of audio
The reason that’s so handy is that there are many transcription and
dictation services where you can email your audio file to them, and they
respond with a written transcript of the contents. Unfortunately I’ve not yet
used any myself, so I don’t have a specific recommendation. I’ve heard of very
cost-effective results using this technique, though, which is why I recommend
it. It’s what I would do should the need arise.
Similarly there are also “virtual assistant” services out there as well that
will do this, and many other types of administrative tasks, whether you’re
traveling or not. Depending on the dollar value of your own time, these can
also be surprisingly cost effective ways to get things done.
And they’ll be significantly less frustrating than jury-rigging your voice
recorder to your PC and getting a transcription full of errors.