Michael Stutz wrote :
>On Tue, 14 Apr 1998, Forrest Cahoon wrote:
>
>> What sound card is the best for recording?
>...
>> I'm thinking of just getting a new card, but which one?
>
>Hehe... been looking for the answer to this one for a looong time... in
>short, I don't think there's an easy answer (but would _love_ for
someone to
>prove me wrong); check out this good netnews posting:
><http://x2.dejanews.com/getdoc.xp?AN=323713887>
>Incidentally, the author of that post is on this list, and has said
that he
>couldn't get full duplex on that card to work with all software.
That's me again! I wonder how long I'm going to keep seeing pointers to
that post?
The problem with trying to get full duplex recording working in Linux is
there are NO free apps I'm aware of that support full duplex with the
stock OSS drivers and a stable (2.0) kernel:
--Ceres Soundstudio works with OSS-free 3.5.x and Linux 2.0.x, but costs
$40 and didn't really seem worth it to me
--Broadcast 2.1 MIGHT support OSS-free and Linux 2.0.x but I can't get
it to work without seg-faulting every time I try to use it *at all*
--Multitrack was designed to be used with two half-duplex cards, not one
full-duplex one; to use a full-duplex card requires the $20 OSS-Linux
drivers (this works well for me).
--SLab is designed for 2.1.x kernels and is too complicated, buggy, and
slow for my taste anyway. Maybe in the future...
>I think the basic problem with sound right now is that the free sound
driver
>that comes with Linux has limitations for which you have to buy the
>non-free, commercial driver to correct. There are some projects to
write
>high-performance--and free--drivers, listed at that page above. But
they're
>all in alpha stages. I've concluded that GNU/Linux sound support is in
its
>infancy.
My experience seems to suggest you're right...
Yea, there's a ton of cool programs but there are big problems with
>almost all of them, as well as problems with the underlying sound
drivers.
Yep. But there are new developments constantly. I think within 2 years
we're going to see Linux as a truly kick-ass environment for working
with sound. For some purposes, it already is... Dave Phillips has been
scraping along with a 486-100 for years now, which is practically a joke
machine to Win95 sound people.
>My solution 3 years ago was to get a GUS PnP Pro, which was a great
card and
>had great support by way of the Linux Ultrasound Project (which is now
>defunct, and the author is working on a free high-performance sound
driver
>for linux). Recently I just bought a TB Malibu which seemed to have a
good
>price/perf ratio (see post above).
Yep, I think it's a great buy for $70 or so, but I did have to spend
another $20 on the OSS drivers to get Multitrack working right. AND I
still get an annoying CLICK every time I start playing a sound. This
doesn't happen in Win95, but it happens in Linux regardless of whether I
use the commercial or free drivers, and regardless of whether I use the
Malibu or the onboard Vibra16 chipset. It's driving me nuts. I think it
must be a driver problem...
As for card bargains, it may be worth checking into how good the
OSS-Linux support is for the Ensoniq AudioPCI which is quite cheap and
is supposed to sound about as good as the Malibu.
>> Special bonus question: what's the best way to generate a sound
sample
>> from an analog source to put on a web page.
>
>When I recorded my 4-track creations to disk, I used "ultrarec," which
is
>basically an ultrasound variation of the standard "brec" program.
Record
>using Red Book CD audio specs: 44.1KHz, stereo, 16-bit, and you should
be
>fine.
I tend to use one or another of my favorite editors (DAP, Snd, or
MiXViews) for recording because I will almost always end up snipping the
beginning/end of the file and maybe tweaking a few other things...
Need LOTS of memory for this. Remember that CD-spec audio eats up over
10 MB per minute, and then you need more memory for the waveform
display... Snd and MiXViews seem to handle big files faster than DAP.
>I put two formats on the Web: mp3, which is near-cd-quality but doesn't
>take up tons of disk space (it takes up a lot, but not as much as raw
cd
>audio);
FYI, mp3 encoding produces a file of 800-900 kB per minute if your
original file is 16/44.1/stereo. You can cut that in half by making the
source file mono or by using a 22050 sampling rate, which may not sound
bad depending on the source sound. For really crunching it, 16/44.1/mono
would end up about 200 kB per minute or so. You could also make the bit
depth 8 instead of 16 but that WILL sound bad, unless you put a lot of
work into maximizing its capabilities, in which case it will only sound
pretty bad rather than bad. :)
> you also have to use a not-completely-free program to encode the mp3
>file.
Yes, it's too bad. I use the venerable l3enc which is at
http://www.iis.fhg.de/departs/amm/layer3/index.html#share
-- you get something like 30 days of free operation, then it
self-destructs... Encoding is kinda slow, especially if you use the hq
(high-quality) flag, but the sound quality is much better than anything
else at comparable compression ratios. It's pretty darn good in fact.
Only big problems I've heard so far are some phase "whooshing" on the
high end, especially on cymbals, and that just has to be tolerated given
that you're trying to put sound on the web...
> The second format I used was RealAudio, which everybody seems to use
>but doesn't take up nearly as much space. I only put a few songs in
>RealAudio though, becuase the sound quality is worthless.
Amen to that.
--PW
Please disregard the evil advertisement below. I will soon be kissing
this #%**)@& account goodbye and getting REAl email...
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com
