Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 30 Mar 2002 01:43:11 +0100
From:      Rahul Siddharthan <rsidd@online.fr>
To:        John Utz <john@utzweb.net>
Cc:        "Adam D. Gorski" <agorski@engin.umich.edu>, freebsd-multimedia@FreeBSD.ORG
Subject:   Re: SB problem (was: Cat'ing /dev/audio)
Message-ID:  <20020330004311.GA1858@lpt.ens.fr>
In-Reply-To: <Pine.LNX.4.44.0203291744280.11448-100000@jupiter.linuxengine.net>
References:  <20020330004107.B88610@lpt.ens.fr> <Pine.LNX.4.44.0203291744280.11448-100000@jupiter.linuxengine.net>

next in thread | previous in thread | raw e-mail | index | archive | help
> i would also concede that your suggestion that the problem is a sampling 
> rate problem is possible. because it would be concievable that linux/alsa 
> might have a more sophisticated approach to automagically resampling 
> things.
> 
> *but*, i simply dont run into content that has been sampled that way! i 
> havent ever run into stuff that i cant play via xmms on my 4.5 box.
> evidently that isnt the case for you.

Is yours the same card (SB16)?  There are lots of cards which accept
variable sampling rates; the problem arises when they don't.  If you
have absolutely no problems with an SB16 whatever the sampling rate
of your soundfile, I'd agree that sampling rate is probably not the
issue in this case. 

Note that with upsampling, you won't get problems nearly as bad.  So
if your card accepts 44100 Hz and all your soundfiles are 44100 Hz or
less, you won't hear the kind of noise Adam complains of; there may be
some slight distortion (that was my experience) but you need good
speakers to make it out.  The problem comes with downsampling, eg
playing a 48000 Hz sound into a 44100 Hz card.  Given that other
people apparently use the SB16 on FreeBSD happily, my guess would be
either Adam's computer clock is too fast (a signal which should be
44100 Hz ends up at 42000 Hz, say) or his cards accept only 22050Hz
for some reason.

And yes, alsa is probably pretty sophisticated about things like
sample rate conversion.  And now that it's become standard in the
linux kernel, more and more linux-based sound applications are going
to require it (some do already) and it will be difficult to port them
to FreeBSD :-(

> my current supposition is a math error, because i know that gcc with the 
> more specialized cpu architecture setups isnt particularly well tested, 
> and it's tested even less on freebsd! furthermore, preparing oggs and 
> mp3's for play is a math intensive process, so any goof ups in the math 
> code would be painfully evident.

Well, gcc3 may well have problems, but the stock gcc (2.95.x) is quite
well tested with vorbis and other audio software.  Also it's not
terribly obvious to me why a math error would affect high sampling
rates and not low ones....

Methinks it's time one of the experts on the list spoke up.

Rahul

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-multimedia" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020330004311.GA1858>