Scott, the answer sort of turned into a status report, so I thought it
might be useful to the list as well--I hope you don't mind:
Scott Rossi <[EMAIL PROTECTED]> said, on 7/4/00 12:52 PM:
>I was wondering if your work with AIFF files would allow for something like
>a script-based audio level meter, which could determine (even vaguely)
>levels of audio in a sound file. Would be neat to build a sort of poor
>man's level meter in MC. Curious what you've found...
I'm not sure; I've used two methods so far in analyzing the sound. In
both cases I grab the sound a thousand samples at a time. It's 16 bits,
22khz, so each batch I grab is about a twentieth of a second. The first
method was to see how many of the samples differed from a base of -500 by
more than x, either positive or negative. I experimented with x from 100
to 300, looking for between 200 and 300 samples that were different by
more than that to indicate that there was noise (that I wanted). Now I'm
comparing each sample to the one previous, and summing the total
differential. If it comes out to more than about 30,000 over 1000
samples, that's noise (that I want). The second method seems to be much
clearer: sounds that I want, even when quiet, total to at least 70-80
thousand, and up to a million when the sound is loud. the background hiss
of the recording never gets above about 20,000.
That said, I don't know why baseline noise seems to hover around -500; is
it my recording setup, and it would be different on another computer?
That's why I switched to the second method, figuring that it was
baseline-independent. The clearer indication of sound was a byproduct.
I also don't know what values would indicate louder versus softer. I
_think_ the method I'm using now, differentials, would give a measure of
that. The only way to find out is to try it.
I'll put up the stack later today so people can try it out. It's very
rough, though... an interface only a hurried programmer could love. :-)
Okay, it's five minutes later, and the answer is yes, sort of. (Ain't it
great?) I now have a scrollbar set as a progress bar, and it sets its
value based on the log of the differential. It seems to work very well,
and hardly slows down the process at all. The sort of comes from the fact
that it's showing the differential on the samples as it reads them in,
which is not the same as the actual sound I'm hearing. I have it going
through a large sound file, cutting out the useful bits, and saving them
each to separate files. It plays each one right after it saves it, but
obviously, the samples it's measuring and displaying can't be the same
ones that are playing at the moment. Still, it's quite impressive.
Again, I'll post it a little later today, and mail the list when it's up.
Thanks for the feedback!
Geoff
Archives: http://www.mail-archive.com/metacard%40lists.best.com/
Info: http://www.xworlds.com/metacard/mailinglist.htm
Please send bug reports to <[EMAIL PROTECTED]>, not this list.
