[Audacity-devel] Re: Audacity-devel digest, Vol 1 #1119 - 3 msgs

[Chris J. <jin...@so...>]

On Saturday, December 13, 2003, at 11:05  PM, audacity-devel-
"Anthony Airon Oetzmann" <ai...@gm...> wrote:
> ASIO please. Pretty please, with shugar on top.
>
>  Tony

	I'm not certain I need it but am inclined to second this request.

	I'm on OSX (finally) and for the
first time ever have been able to record off my M-Audio Audiophile
2496's SPDIF input to disk, on my aging machine. I can only assume
someone in Audacity is running this card as it seems to be not only
handling the input but even doing it at high resolution (I'm recording
to the 32 bit float format and I think I'm getting the additional bits
on the SPDIF generated by a Midiman 'SAM' format converter in mix mode.)

	I'm very psyched- but it seems as if I might be jumping to conclusions 
and just getting a very nice clean 16 bit signal to the 32 bit float. 
Anybody know what Audacity does with this card's digital input if you 
ask it to do 32 bit float?

	I'm also beginning to look into perhaps implementing some or all of
the functionality of my 'Mastering Tools' code into my own version of
Audacity or some sort of plugin. I'm not sure quite how well I'll be
able to do this and suspect I'd be useless at propagating the code to
all supported platforms, but on the other hand it is out there
available and what I lack as a bare-metal coder I make up for in
algorithm design. I've got:

	-transient enhancement in two varieties, one more trebly and one more
midrangey
	-slew limiting and a type of high-loudness peak limiting that can put
out more apparent loudness than Waves L2 on bright material (I don't
myself abuse this but it can be pushed awfully far)
	-a whole history of unusual wordlength reduction approaches
culminating in a particularly unusual one that turns quantization noise
into a combination of sidechain compressed music detail and quiet
intermittent (not continuous) noise. Are you familiar with the dither
shootout at 2496.net? My entries in that did OK but are out of date.
This is what I use now: http://www.airwindows.com/studio/fullcrank.zip
(same test file as used on the dither shootout, a recording dropping
way below quantization noisefloor and then amplified up to top volume
again)
	-a very elaborate metering system depicting peak, stereo RMS, limiting
and slew limiting in a continuous display that could be implemented as
a double buffered scrolling display using a trick from sprite game
programming
	-an unusually 'loose' deep bass boost capable of going positively
subterranean without exceeding excursion limits of woofers
	-a series of mid/side stereo enhancement routines that includes the
above boost on summed content only (wise for effective bass
management!) and a double-duty ambience enhancer and elliptical EQ on
stereo content.

	I'm running on OSX Jaguar (cannot install the OS upgrades though,
because I get kernel panics) and looking forward to seeing where
Audacity takes me. I can think of only one drawback to implementing
this stuff within Audacity: I'm currently running a double precision 64
bit floating point internal buss in Mastering Tools, and I'd be losing
accuracy to go to 32 bit. However, it looks to be well worth it.

	Wish me luck as I begin to grapple with this stuff! I'm no C coder and
have been implementing these algorithms in a friendly toy-language.
I'll see if I can grow beyond that- once I did succeed in making an
early version of the slew limiter as a VST plugin, because I had
example VST plugin code that would build. I can generally translate
into C, but get snowed if the project starts to produce build errors
that aren't directly linked to some obvious coding blunder on my part.

	Chris Johnson