From:	Shawn Hargreaves <Shawn AT talula DOT demon DOT co DOT uk>
Newsgroups:	comp.os.msdos.djgpp
Subject:	Re: Help needed: Allegro outputs WAVs with noice
Date:	Tue, 28 Oct 1997 19:50:19 +0000
Organization:	None
Distribution:	world
Message-ID:	<UBWWEOA7HkV0EwiI@talula.demon.co.uk>
References:	<62qa8s$dr3$1 AT gts DOT nsk DOT su>
NNTP-Posting-Host:	talula.demon.co.uk
MIME-Version:	1.0
Lines:	37
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp

Oleg Shamshura writes:
>I have some WAVs of high quality, created by a professional sound 
>studio. Windows'95 plays them very clean, but Allegro adds noice.
>I have SB AWE 32.

Allegro is using a CPU function to mix several samples into the output
audio stream, to let you play many different sound effects at once. This
loses quite a bit of quality compared to something like the Windows
audio drivers that just dump a single waveform directly to the
hardware...

In Allegro 2.2, there were 8 voices playing at once. In more recent WIP
versions this can be adjusted anywhere between 1 and 32: lowering the
number will slightly improve the quality. But the fact of doing this
mixing means that the output will never be totally clean. Some bits are
lost because I have to reduce the sample range to make room for multiple
voices, and others because the lookup table I use for volume adjustments
would get prohibitively large if I used a full 16 bit waveform! Also,
since Allegro may be playing many sounds at one time it has to adjust
them to whatever frequency the soundcard is running at, while I suspect
Windows will be able to just set the soundcard to the same freq as the
sample: this frequency conversion can introduce some noise since I don't
do it properly with all the interpolation and filtering that a sample
editor would use :-) This all means that Allegro is unable to give you a
full 16 bits of sample precision, even on a 16 bit card (it's better
than 8 bit output, though...)

If you really need to output a waveform without any modification at all,
you can hack the _mix_some_samples() function (this is found in sound.c
in Allegro 2.2, or in mixer.c in WIP versions), and make it just copy
your waveform across into the DMA buffer without any of the mixing code
being used...


--
Shawn Hargreaves - shawn AT talula DOT demon DOT co DOT uk - http://www.talula.demon.co.uk/
Beauty is a French phonetic corruption of a short cloth neck ornament.

- Raw text -