NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/comp/sys/ibm/pc/soundcar/7527

Internet Message Format | 1993-01-28 | 4.5 KB

Path: sparky!uunet!UB.com!pacbell.com!ames!olivea!charnel!sifon!CC.UMontreal.CA!dionf
From: [e-mail] (Francois Dion)
Newsgroups: comp.sys.ibm.pc.soundcard
Subject: Re: ASL Cyber Audio Card - So
Message-ID: <[e-mail]>
Date: 28 Jan 93 16:20:27 GMT
References: <[e-mail]> <[e-mail]> <[e-mail]>
Sender: [e-mail] (Administration de Cnews)
Organization: Universite de Montreal
Lines: 79

In article <[e-mail]> [e-mail] (david cole) writes:
>In article <[e-mail]> [e-mail] (Francois Dion) writes:
>>But only the GUS has 1 Meg ram and not ROM. You can put around 40-50 GUS
>>patches in 1 Meg, so 128 sampled instruments in 1 Meg ROM is a really
>>bad idea.
>>If you put more patches in the same area, you have lower quality samples.
>>To hold all the standard GUS patches you need 4 Meg.
>
>This assumes that the samples in ROM are not compressed.  But there is no

Just to get the equivalent of what the GUS has you would need 4:1 NON lossy
compression. I'm very surprised you said that, because i think you read
comp.dsp, so you should know that it is impossible. A sample of instrument
(like those for the Gravis) is really hard to compress by more than 15% with
a non-lossy algorythm. With lossy algorythms you get better, but at 4:1 you
will get very noticeable effects (you really want those patches as clean as
possible). Ask anybody on comp.dsp, dat-heads, the MPEG group, etc...

>basis for that assumption.  When I have seen descriptions of synths that use
>ROM for waves, they often mention that the waves are compressed.  Since the
>hardware is dedicated and the format is proprietary, there is absolutely no
>reason NOT to use compressed samples.  Since sound is highly redundant, the
>compression must be considerable.

Sound is not redundant at all. To get a good start, read the comp.dsp FAQ.
Not only that, but the real GM will probably take 5 meg, since several patches
lack presently. (Just all the drums take 1 Meg)

I said:
>>Another advantage of RAM is that it allows MOD players of 1 meg mod files
>>without using a dos extender, and using only 64K main ram.

And not only that, but the sound quality is much better this way.

>>A third advantage is the fact that you need RAM to use FOCAL POINT 3D audio
>>technology to preprocess your samples. (which is exclusive to the GUS on the
>>IBM platform for 1 year, with option for 4 more years)
>>
>There is other 3D audio technology around.  I believe Roland has demoed it.
>I would be surprised if it requires samples to be in RAM, since anything in
>RAM can be in ROM.

Read my 3d.txt description of all the 3d methods on archive.epas.utoronto.ca.
The RSS has been around since 91 (Roland Spacial Sound) but it requires a
costly DSP to do it. Not only that but you then need to implement controls by
registers etc... Presently it is only available for a huge amount of money,
and must be controlled thru midi. It will take several years before it is
implemented in a soundcard (minimum 2 years, and even then it will be expensive)
so Focal Point is the only choice for now, exclusive to Gravis.

>>A fourth advantage is that you can use any sound and sample it and put it on
>>the card (be it a Juno-106, a TR909, a pan flute, a chainsaw etc...)
>
>This is undeniably the major advantage of using RAM.  Seems to me the best
>card would use both ROM and RAM, like high end synths do.  If there is a
>point to putting stuff in RAM, it is to avoid loading from a disk and to make
>sure there is a reliable standard set of samples available -- the Gen Midi
>set for example.  I don't see any point in having the Gen Midi set in RAM.
>It just makes the process slow and unreliable.  Save RAM for custom sounds
>and upgraded sounds.  4 meg ROM and 4 meg RAM seems like a reasonable size
>for a quality card -- and onboard reverb of course!

The GF1 doesn't support that in its present form. I think Gravis was caught
by surprise that everybody wanted a 1 meg card, and that if they had sold
a card with the possibility for 4 meg, people would probably have upgraded
to 4.

>Cheers,
>
>Dave

Ciao,
--
Francois Dion
    '  _   _   _
 CISM (_) (_)  _) FM       Montreal , Canada       Email: [e-mail]
      (_)  / . _)             10000 Watts          Telephone no: [phone]
_______________________________________________________________________________
Audio-C-DJ-Fractals-Future-Label-Multimedia-Music-Radio-Rave-Video-VR-Volvo-...