Tim Hall wrote:
> The difference is that a
> synth that uses samples uses relatively short bursts of sound, mostly the
> attack portion, that the ear uses to differentiate instruments and various
> loop portions, the difference is made up with synthesised sound.
> This kind of thing exists in soundfonts, usable by fluidsynth and editable
> with smurf/swami (in theory)
The term "wave table" refers to looking up the sequence of values in a
table or an array, not where the samples came from nor how they were
created. A wave table synth is a type of sampler synth. There
is no requirement that the violin begin with a sample of the attack of
a real violin, then transform into an oscillator. This could be done,
and the values could be stored in a table, subsequently played out
a wave table synth. But this doesn't determine that such a synth
has the name "wave table synth." One can record knee slaps and
put them into a soundfont, and then play them out their SB Live! card
with no synthesis anywhere in this process. Longer notes of extended
samples can be created by looping back through the sample. One doesn't
need to resort to synthesized sound for the remainder of a note, yet
still legitimately refer to their synth as a "wave table synth."
> a sampler accomplishes basically the same
> thing as a wavetable synth -- it uses sound samples to generate
> tones, doing frequency shifting and interpolation as necessary.
I would recommend considering a wavetable synth to be a type of
> And as I understand it, the main difference between a sampler and
> a wavetable synth is the lack of constraints on the samples used
> -- with a sampler, anything at all could be a perfectly good
> sample, including samples of almost arbitrary duration (and thus
This is not really correct, but an implementation detail.
ALL sampler synths, including wavetable synths, have limitations
on the samples. Now sometimes you'll see marketingspeak: "Limited
only by the capabilities of your machine." There may be no hard-
coded limits, but there are indeed limits.
> But that brings my
> first question -- if you don't own/play the instruments in question,
> where do you get the samples? I've done a lot of web searching,
> and found tons of drum loops and bass lines that are two measures
> long and so forth, but don't find much in the way of e.g. individual
> notes on basses.
When someone asks "Where does one obtain samples," many immediately
advise the questioner to go to the many sample libraries which exist,
sample CD's, sample loops, the Internet sources, etc. One can also
record one's own samples. One can also simulate instruments through
physical modelling. One can also record one's own samples of
*whatever can be sent through the audio path*. One can mix samples,
including individual notes. For masochists, one can type in a table
of values, convert this to an audio format, then use that as samples.
One can "rip" them off CD's, videotapes, radio broadcasts (the legality
of which depends on the source and the use of the material). There
are *many* sources of samples. The main limitation is the composer's
A simple example: Record yourself humming into a mic. Go into
a WAV editor and give this a guitar envelope. If you can, build an SF2
font. Now play some guitar melody. Send it through an effects processor.
Those who have never done anything like this will learn a lot. Now
repeat with some other sustained sound... In some wave editors you can
keep just the envelope. So take a spoon and tap on something. Keep
this envelope and apply it to your humming or other sustained sound.
Rip a track of a CD and isolate one sound that you really like that is
a solo part, and edit it down to one second. Apply the envelope to
this. Build a font and play something. (Specimen by Pete Bessman
can be used to do this sort of thing without having to build fonts.)
> you're gonna be
> spending hours and hours trying to find samples that work.
In my opinion, this is always true for great music. You can either spend
years learning how to play an instrument, or you can spend years learning
how to sythesize, record, or manipulate samples. Otherwise your stuff
sounds like everybody else's and/or you begin to repeat yourself.
Sorry to be a pest, but does the Hammerfall Lite fall right into
the box on Linux? If I'm seeing right there is not the possibility with
this card to have playback without a computer since it is nothing but
optical ports, with no 1/8" or 1/4" jacks? This Lite card is what I
have been leaning towards but want to be really sure about it because I
am researching a whole new setup based around a good optical card.
Just out of curiosity, what preamps and such do you interface to this
You said that you didn't use the Lite card much under Linux but
that it worked ok, and sorry to be a pest here, but were there
deficiencies somewhere with this card that caused you to yank it out?
>> I also own a Hammerfall Light. It worked under Linux, but I didn;t
use it much. Some time ago I removed it and it's sitting on my desk
begging for a machine to go into.
That wouldn't be a sales offer, would it? ;>)
Many thanks again,
NOTE: I am cc-ing this to the kernel list in hope someone there might have a better insight in this. For the kernel people who intend to respond to this, I would greatly appreciate it if you could CC me, as I am not subscribed to the kernel list.
snd-hdsp (RME Multiface cardbus pro-audio soundcard) works in Linux but the sound is trashed (distorted). In Windows on the same notebook, everything works fine. The problem has been now reported on 2 completely different notebooks (Acer 1400 with 02 cardbus and my eMachines m6805 with ENE CB1410 cardbus controller). I suspect at this point that the culprit is most likely the cardbus driver (yenta_socket in my case).
After digging some more I am absolutely confident that my problem with the hdsp_multiface+laptop(cardbus) is definitely not only similar but identical to the problem Tim Blechmann reported in January. The scary part is that my laptop is completely different brandname than his and uses entirely different cardbus (IIRC he has Acer 1400 with the O2 cardbus; I have eMachines m6807 with ENE CB1410 cardbus).
I think that this now has to have something to do with the current state of the kernel cardbus drivers (pcmcia-cs has not been updated since December so I would assume that they are no better than the ones that are found in the kernel) and possibly the updated hdsp driver (although not entirely sure on this last one).
Here's the current scoop on the problem:
Windows XP -- stuff works great, everything as expected. The only thing is that when the computer goes to standby/hibernate, upon resuming the sound is all distorted (just like Tim reported it -- slower, full of artifacts, but you can still discern the original sound's content); this most likely has to do with the crappy BIOS my notebook has (esp. in respect to the ACPI and APIC -- DSDT table is trashed etc.). After distortion occurs, overclocking the computer seems to speed the sound up bringing it closer to the desired playback speed but the artifacts remain. Miller Puckette suggested that perhaps the hdsp is not getting proper clock info from the CPU, something that I have not investigated as of yet as I do not currently have access to an external equipment that would provide Word Clock functionality. Although this also sounds a bit weird as the soundcard works just fine upon first boot (prior to suspending the computer). No matter how many times I reconnect the card and/or mess with it before suspending the computer, everything continues to work as expected.
Mandrake Community 10.0
Plenty of RAM and other junk
IRQ for cardbus and hdsp is shared on 11 (them sharing the same irq IIRC should be normal behavior)
Alsa 1.0.2 and 1.0.3 tested (1.0.2 came with the system, 1.0.3 compiled from source)
Latest Jack and alsaplayer packages compiled from source
Hw:0 onboard via82xx
ACPI and APIC are disabled due to BIOS issues with the laptop and because even with the pci=noacpi flag in lilo the system still freezes when inserting cardbus. I saw somewhere a kernel patch that would enable use of cardbus with a limited acpi presence (pci=noacpi) but have not tried using it just yet mainly since presence of acpi should not have any positive bearing on resolving this issue (if anything, it would make it even worse due to IRQ shuffling).
Modprobing goes without a hitch, pcmcia service automatically starts, the cardbus interface using yenta_socket driver. Snd-hdsp also works without a hitch and configuring the soundcard is all ok (hdsploader, hdspconf, hdspmixer all check-out fine).
aplay D plughw:1 (or plughw:hw:1 forget the exact syntax currently booted into Windoze) <soundfile> plays the sound distorted similarly like in Windows after resume.
Jackd d alsa d hw:1 with various flags and sampling rates of either 44100 or 48000 works without any dropouts. Just like with Tim, no distortion is coming through until the sound is played. During the sound playback, the distortion is identical to the one during the playback without jackd.
Connecting simple clients like jack_metro plays stuff, but distorted.
Alsaplayer when connected via jackd also works but again distorted. I have to put the playback at 200% speed to get the right tempo of the song but the sound of the songs singer is now very high-pitched (chipmunk?). Distortion persists no matter what.
Reconnecting cardbus and all that works but the distorted sound persists.
Hdspmixer shows the sound levels as expected and they reflect the fact that the sound is being played slower than it should be and that it is distorted.
Messing with hdspconf during playback makes no difference. Adjusting the sampling rate though does alter the sound of distortion when playing (just like in Tims case), but does not alleviate it.
/var/log/syslog lists no complaints and/or problems. (I will check more thoroughly for the boot-time stuff).
I am aware of the fact that the BIOS is somewhat trashed (manufacturers fault) but not to the point where machine does not behave normally, esp. in Windows.
Some have suggested switching distros, but my understanding is that Tim was running gentoo and I am running Mandrake and were both having the same problem
Tim, I believe also used 2.4 kernel without success, as well as pre-1.0 Alsa drivers.
1) provide detailed lspci
2) thoroughly check /var/log/syslog for anything suspicious
3) try pcmcia-cs (most likely wont work as Tim already tried that and it made no difference on his laptop, also the package wasnt updated since Dec.)
4) try playback with an external Word Clock source
5) provide downloadable examples of the distorted sounds
6) Pester alsa-dev, lau, and kernel/pcmcia people to death begging for help :-)
7) Pester eMachines to update BIOS (I may retire before this one happens, though)
8) Something else?
I would appreciate any help with this one especially now that we know that the problem is not related to one particular notebook/cardbus controller
I will provide additional info as soon as I get home (sometime early next week).
Ivica Ico Bukvic, composer & multimedia sculptor
Thanks for this valuable input. So...if the HDSP9652 is not the
best ADAT card available for Linux, would the Digi9636 (the HammerFall
Lite) be a better option? I found a discussion of this very thing from
earlier in the month, but am wondering why the AP2496 limits your
options for real-time recording work? The AP2496 has no optical inputs
and the DiO 2496 from M-Audio seems to be the only card they have that
does have them, and that is only 2 X 4 in/out. I don't seem to have
many options in choosing ADAT cards here.
Mark Knecht from earlier March 8th post:
>> Don't get the stand alone HDSP9652 or HDSP9636 until you make sure
the Linux drivers will do what you need.
You had said this on a previous post and I was wondering what you meant
by stand alone. Which RME card seems to work the best for recording
work? I'm looking at that new M-Audio Octane as a possibility. Am I
correct in that the RME cards have the converters on them, or are the
converters on the preamp/optical out interface?
Thanks for bearing with me on this,
Last nite I read that article in SosPubs about Linux. Most
folks in that article seemed to be using the RME HDSP 9652. I have been
looking at the MultiFace for awhile, but am starting to feel that the
best option would be to sink my money into some of the preamp/ADAT
optical out outboard gear that is starting to come on the scene such as
the M-Audio Octane, Tango24, or Behringher ADA8000 that feature lots of
preamps w/ ADAT optical outputs, and just using the HDSP9652 as a really
nice signal router into my Linux box. Would anyone be willing to share
with me the pros and cons of this approach?
Up until now I have been thinking about sinking all of my money
into the MultiFace or a really high-priced audio card with minimal
outboard. Now I'm starting to think just the opposite that just using a
bare minimum optical card w/ decent converters and putting my money into
outboard would maybe be the best way. Any good advice on this, and if
so, any other cards I should consider?
Thanks very much,
Hi. I've been playing guitar for 7 years or so, and using linux in one
way or another for about the same amount of time; but only this past year
have I started playing around with what I can do musically with my linux
One of the things I've recently learned about is samplers. As I
understand it at this point (and I'm hoping someone will set me
straight if I'm wrong), a sampler accomplishes basically the same
thing as a wavetable synth -- it uses sound samples to generate
tones, doing frequency shifting and interpolation as necessary.
And as I understand it, the main difference between a sampler and
a wavetable synth is the lack of constraints on the samples used
-- with a sampler, anything at all could be a perfectly good
sample, including samples of almost arbitrary duration (and thus
One of the most obvious uses I can see for a sampler would be to
use it to provide instrumentation that the user doesn't know how
to play. For instance, if I wanted to record myself on guitar
with a piano accompaniment, I could use a sequencer to write the
piano line and generate it through a sampler. But that brings my
first question -- if you don't own/play the instruments in question,
where do you get the samples? I've done a lot of web searching,
and found tons of drum loops and bass lines that are two measures
long and so forth, but don't find much in the way of e.g. individual
notes on basses.
And I wonder about how people use the extended samples I find.
It seems kinda constraining, to be stuck with a melody/harmony
line given to you by whatever someone sampled. Of course, there
are tons and tons of samples available; but then, in order to
express the music you're hearing in your head, you're gonna be
spending hours and hours trying to find samples that work.
Am I missing some obvious things here? How do people use samplers,
for the most part?
Thanks for dealing with my beginner-type questions.
Chris Metzler cmetzler(a)speakeasy.snip-me.net
(remove "snip-me." to email)
"As a child I understood how to give; I have forgotten this grace since I
have become civilized." - Chief Luther Standing Bear