Print

[usrfriendly]

Oops... skipped Kmuland's post (sorry, man).  Got to Jojo's(?) and skipped.  I figured that's how it's done, but how do I know what note it's tuned to.  My tuner (chromatic, btw), only picks up the note after it's done playing.  I'll give it a try tomorrow.  Thanks, all.

[TraumFlug]

tuning samples properly may seem senseless to people who just want to create tunes and listen to them one after another, but imagine following examples:

- a dj wants to mix 2 tunes in sync, so he better makes sure both tempo and tuning of both tracks match in a way that makes them sound good together. modern dj mix tools might have pitchshifting/timestretching to make that task easier, but many are proud of their skill to match songs with pure replay-speed adjustments

-you have a song, and want to include recorded tracks, like a guitar solo or whatever. though the samples can get very long, and it really bloats up an .xm, it's possible. I've already done this, and it works well (unless the sample is very long, and the tracker doesn't check for sample size, causing it to crash due to some overflow - I've had that with jeskola buzz, making a not-backuped song impossoble to load, because saving the .bmx seemed to work) only problem is that the track can't be changed in speed, unless you use a timestretching tune on your recorded parts.

the best sollution is to create an instrument with a simple beep as looped sample. probably a sinewave, or whatever you fancy, and make sure it's "A" is tuned to 440 Hz. save it as "tuning.xi"
and load for every song you build, tuning each new instrument to that sound. I could imagine making a pattern where this beep is triggered in the beginning, in the pitch you think works best, or that is appropriate because you have a multisample instrument. now you let that pattern loop and "keyjazz" the same note in the instrument/sample you want to tune. you need good ears for that, it's similliar to tuning a guitar without help of a tuning gadget: if you think the result sounds "wobbly" or "spacy" a little, i.e. pulsating readjust until the pulsing gets slower. the slower it is, the closer both sounds are together; in ideal conditions you won't hear any chorus-like effect, because both sounds have the same base-freq.

[dysamoria]

another good reason to tune your samples is... if you have a lot of samples in your collection and you've written your song and are working on sound design and swapping out samples/instruments to find better choices, writing a song with "incorrect note values" leads to having no baseline whatsoever when you swap sounds out. if a self contained song is all tuned to itself, that's great, but when you want to trade sounds it becomes a problem if the tuning is off between what you have already and what you want to pop in. do i fuss with the sample tuning and retain a bunch of note values that done truly represent what is being heard or do i modify the notes? modifying the notes in a song is a huge structural change and makes instrument/sample swapping even harder. so tuning instruments/samples so that they're all properly indicated as their correct audio pitch is a good idea.

keeping a library of samples that are in tune with each other is useful for efficient music work.

now... HOW to do it... well that's the complicated part. one thing to do is make sure you're always using the same sampling rate. 44.1KHz is the "standard" for human hearing baseline and audio playback on computers (loosely speaking). if all your samples are 44.1KHz, the next thing to worry about is the content of the recordings themselves.

the content: many sample format files have a "root note" saved in them. this tells programs (and users) what the actual pitch of the thing recorded was. this is how you match up what you hear with what you see on screen. you can change how they sample is played back when inserted into a sample player (tracker or otherwise), but that doesn't help when you have a lot of samples all varying and you're trying out new things. keeping things in tune from step one is important. if you have to make slight tuning modifications to samples, you can use audio editors like Sound Forge to shift pitch/tuning without resampling the file or changing its playback rate.

overall, i advise comparing your sample to the same note on a piano, keyboard synth, etc and tuning your sample to match. the guitar tuner is a great idea, but it's not so useful with complex timbres. we have to rely on our ears a lot.

when you make recordings, ensure whatever you're sampling is either in tune, or you save the file with tuning info (root note, note tuning, etc) so programs that use it can have a sanity check on what the content is supposed to be frequency-wise.

the previous post above mine, about loading a tune sample for the whole purpose of matching other samples to it is brilliant and i wish i'd seen that tip ages ago!

i hope some of my rambles are useful.

[DasKreestof]

My problem is with the actual tuning of the instruments.  I have one tuned to C-4, one at C-2, both were sampled that way, but my third sample, after adjusting the loop points, came out at F#-3.  I'm trying to figure out how to tune that sample to the others.  I haven't gotten any change.

Seeing the F# in this question made me wonder if perhaps the real question here was about how importing samples with a sample rate of 44.1khz changes the relative note of an instrument to F6 fine tune -28.

I admit that it took me a while to understand what was going on. It's best explained in a few other threads on this forum, such as this one: http://modarchive.org/forums/index.php?topic=1195.0

To sum it up, most of us from non-tracking backgrounds in digital audio are used to working in 44.1khz or 48khz (or some multiple of those) environments. Milkytrackers base environment is 8363hz, so a middle C sampled at 44.1khz, must have it's relative note altered F6 fine tune -28 to play it back at 44.1khz instead of 8363hz.
The abstract idea is that the wav file is just pixelated wave data. The speed of playback is chosen by the playback system which in this case is determined by the relative note. A relative note of c4 ft0 will play back that data at the rate of 8363. If your sample was sampled at 8363hz, than c4 ft0 will be perfectly tuned. (For example, if you generated your tone in Milkytracker)

For example, say you sample A4 twice.
Sample A : recorded at 44.1khz
Sample B : recorded at 8363hz.

If you play those two samples back at their respective sample rates, you will here the correct tone of 440 cycles per second.

Now import those two samples into Milkytracker.
If you play those two samples with the same relative note, they will be very different.
Upon import Milkytracker will recognize the correct sample rate for Sample A: and will adjust the relative tuning so that Sample A will be played back at F6 fine tune -28 so that upon playback it will play at the correct rate.

I hope this is helpful.
____
Edited: replaced F#5 Fine tune +36 with F6 fine tune -28. I just did an import and that's what I got. I'm not sure where I got the F#5 Fine tune +36 from.

[dysamoria]

yeah, this is very useful info. thanks for the a/b examples, too. i had no idea that MilkyTracker and older trackers used this base environment of 8363hz. is this something Amiga specific? i have a Tandy 1000 TL/2 with a built in 3-voice digital audio chip (wonder if anyone ever made a Tandy specific tracker) and the default sampling rate for the software bundled with the computer's Deskmate environment is 11,000Hz.

to complicate things more, i cannot edit or even play samples in Sound Forge with ASIO enabled bc ASIO doesn't support anything lower than 44.1KHz. i have to use Windows Direct Sound driver, which doesn't like these low sample rates either. After playback (which is sluggish on triggering, maybe because i'm on an old P3 500MHz PC), there's a glitchy noise (likely the playback engine or SBLive Kx driver is futzing around resetting default frequencies or something, kinda like a DC offset makes clicking... does that make sense?).

i could use the MilkyTracker sample editor, but Sound Forge is my personal favorite "native" familiar environment & is of course more advanced in general than an integrated sampler in any tracker.

it seems that to use a tracker, especially one designed with legacy support (which is a good thing), requires knowing a lot about what that legacy is, how it works and why, and then staying within those confines very carefully.

[DasKreestof]

Yes, the 8363 is a legacy rate based on Amiga Hardware. It's mentioned in a thread here somewhere.
You don't need to downsample your samples to 8363. Just allow to Milkytracker to automatically adjust the tuning to F6 fine tune -28 for 44.1khz samples and your samples will play back in tune.
In theory, a Middle C sampled at 44.1khz played back with a relatvie note tuning of F6 fine tune -28 should be in tune with the same middle C sampled at 8363 tuned to middle c.

You could test this with some generated Pulse waves imported into Milkytracker and then exported to Wave. "A" notes are the best for counting since they are usually whole numbers.

As for your ASIO drivers, perhaps a driver set like "ASIO4all could use work with alt sample rates.  I'm not sure, perhaps you're stuck using the WDM driver when using these rate. I'm a cooledit guy which has pretty flexible with sample rates. (perhaps it's coded to recognize that the playback rate of the sample has nothing to do with the audio rate to send to the card?
I agree that for serious sample editing, dedicated sample editors are always going to be more powerfull.

I have an sblive and haven't run into the issue you're having. Perhaps there's a setting in Soundforge you need to mess with.
I don't think the p3 500mhz pc is a factor. I think it's probably a config setting for how soundforge works with the asio driver.
___
Edited: replaced F#5 Fine tune +36 with F6 fine tune -28. I just did an import and that's what I got. I'm not sure where I got the F#5 Fine tune +36 from.

[dysamoria]

ASIO4ALL was considerably slower than the plain DirectSound driver mode (this was when using the Creative drivers). i couldn't use ASIO at all without horrible dropped sample glitching.

 i changed over to the non-Creative kX driver to see if it was more efficient and it turns out to be much better. in ASIO mode, it provides me more "tracker power" with less CPU use. i recommend them if you need more efficient drivers. http://kxproject.com/

Sound Forge and ASIO... i think their ASIO support is limited to 44.1KHz and up (nothing lower). it complains that the driver is unable to support the sample rate/bit depth. when i switch to DirectSound (instead of ASIO) it plays, but has that odd noise after playing is finished i mentioned earlier. i assume ASIO & Sound Forge are simply not designed with the intention to deal with such sample rates since they're not considered modern or "valuable" formats. but MilkyTracker, Awave Studio and Renoise have no glitching on playing these low-rate samples (even with Renoise and Milky Tracker using ASIO - i assume they're mixing at 44.1KHz, where Sound Forge tries to play a sample at its native rate and then runs into driver issues). so, all is really ok in the end.

as for tuning... i have a neurological impairment that makes calculation/math and other number stuff hard for me. ironically, i'm a tracker user... hah... (i'm better at using references for effects than remembering the effect codes, and i'm not very good at utilizing them like really good composers, but i get by). so... i'll keep plugging away at it and maybe figure it out with the info you've provided and experimentation.

[DasKreestof]

I always thought math was going to be helpful later in life, but I never imagined it would have been music where I'd wind up using it so often.

That said, you can let Milkytracker do all the math for you.

If you're interested in scientifically determining how precise the tuning is:
If you want to experiment with tuning imprecision the basic idea I want to present is that when you generate a note A square wave in the fourth octave, the length of the single cycle should be 19.0068 samples at 8363hz aka (440/8363) or 100.2272 at 44.1khz(440/44100)

If you took those two samples of the same note, but recorded with different sample rates  and brought them into Milkytracker the 44.1khz sample would automatically have it's root relative note adjusted to F6 fine tune -28. (milkytracker is doing the math for you.) If you play back the two notes one after the other and export it to wave, you can count the lengths of the square waves of both notes and determine how different they are from one another.

Musically, it's should be close enough that it doesn't matter. But if scientifically interested, this would be a good testing methodology.

I choose square waves because it's so easy to determine where the oscillator begins and ends.
____
Edited: replaced F#5 Fine tune +36 with F6 fine tune -28. I just did an import and that's what I got. I'm not sure where I got the F#5 Fine tune +36 from.

[dysamoria]

yeah, i never thought i would use math... then i realized i'd been doing HEX for ages in trackers and most people in my math classes in school never even knew what the heck that was. decimal was all they knew and didn't know what it was or why it was. hah. but my limitations are more along the lines of mental processing and measurement. i have disabilities such as dyslexia and dyscalculia (really exists).

does MilkyTracker do anything for an audio file that's at a sampling rate of 30,000Hz? (any compensation, i mean). or does it compensate for ANY sample that's not at its base rate? if it does compensate for all files, then my problems are more about how to USE the sample and instrument editors to define the root note and then map the samples across the keys correctly based on that root note.

[DasKreestof]

I don't know how Milkytracker treats other imported sample rates.
It's possible that sample editors (such as sound forge) might write different header data than instrument editors such as Awave.

If I was you, I would just experiment and find out.  When you load a 30khz wav into milky, does it alter the root note and fine tuning?
If not, take a sample of a square wave taken at 30khz.
And then see how you need to adjust it for it to play back correctly in Milkytracker. 
To do this, I'd adjust it mostly by ear against your computer playing it back at 30khz. And then to do the final adjustments, play back the note in a Milkytracker song and have Milkytracker export it to a 44.1khz wav. Count the sample in a single Oscillator. If the osc is longer or shorter than it should be, adjust the tuning. If it's as close as the fine tuning will allow, then you've found the proper tuning for 30khz.

Unless someone else knows the answer and wants to pipe in.