NUSofting

WilliamK · Post by **WilliamK** » Mon Jul 28, 2003 7:11 pm

Thanxs guys, just remember that Luigi is onboard too. We are also fixing Combo and Hornet with the stuff we found while testing TubiLeSax. Isn't coding great?

Wk

McLilith · Post by **McLilith** » Mon Jul 28, 2003 7:16 pm

Hi William,

I just wanted to ask if you got the message I sent you about TubiLeSax and intonation problems. Just in case your in box was full or something, I could either re-send it or post it here. (I don't expect you to have a solution yet, just checking to see if you got the message.)

thanks,
McLilith

WilliamK · Post by **WilliamK** » Mon Jul 28, 2003 7:48 pm

Sorry McLilith, I did get the messages, and I saved both. I can't do all this right now, since they are not actual bugs.

But I will keep that in mind. Thanxs.

Wk

McLilith · Post by **McLilith** » Mon Jul 28, 2003 8:14 pm

Sorry McLilith, I did get the messages, and I saved both. I can't do all this right now, since they are not actual bugs. But I will keep that in mind.

I know the first message I sent was not regarding a bug, and I pretty much expected you to be too busy to try and implement that idea at this time.

However, regarding my second message, are you saying that the significant tuning errors in TubiLeSax are not "bugs" or at least something to be concerned about at this time?

If the tuning errors are basically a design decision to emulate a physical instrument that has tuning issues of its own, that's fine. However the tuning errors in TubiLeSax are rather large. (Some people would call this a bug.)

Just so everyone know's what I'm talking about: If you play two instances of TubiLeSax in unison, and set one Tubi to "Alto" sax and the other Tubi to "Tenor" sax, they don't match each other in intonation. Some notes are very out of tune, while others are pretty much in tune. You can try it yourself, and decide if it would be a problem for you. When using the sax for a solo part, it isn't so much of an issue, but there are pieces of music that call for more than one sax, and this might become a problem for some people.

Also, you can get similar errors by trying to match "Alto" with "Alto II", or "Alto II" with "Tenor".

I just want to help make sure that Tubi is the way you intended it to be. I hope I don't sound like a nag, because I know how it feels to be in your situation, and I certainly wouldn't want to annoy you.

take care,

McLilith

WilliamK · Post by **WilliamK** » Mon Jul 28, 2003 8:21 pm

The only thing out tune could be semitones/cents but not whole notes or more than 0.1 cents. If you do find something totally out of tune, tell me the note and SoundSet. Maybe I did something wrong and I didn't hear.

Also, to be sure, I'm running my tuning on all notes again.

Wk

WilliamK · Post by **WilliamK** » Mon Jul 28, 2003 9:06 pm

ARGHS I just found the tune problems. I tuned wrong some of the files. I'm fixing it, sorry about that.

Wk

McLilith · Post by **McLilith** » Mon Jul 28, 2003 9:22 pm

ARGHS I just found the tune problems. I tuned wrong some of the files. I'm fixing it, sorry about that.

I'm glad it wasn't a case of me just being overly picky.

I think you'll find a high "E" note that is particularly shrill if you look around enough.

Even with the tuning issue, it's still a nice VSTi and you've done fine work on it so far. It should be really great by the time it gets released. I can't see myself wanting "Candy" when this is finished, and I bet a lot of other people will feel the same way.

thanks,
McLilith

WilliamK · Post by **WilliamK** » Mon Jul 28, 2003 10:02 pm

NEW UPDATE 0.91

We just uploaded the FULL SampleSet again, we managed to find all tune problems and also did better LOOP points (as Ford pointed) but is almost impossible to do perfect loop points with the samples we got right now. If the sales of the final product woes we can go and record more samples at the studio.

I also put back the Vibrato code, it was commented for prev tests, sorry.

Regards, WilliamK

McLilith · Post by **McLilith** » Mon Jul 28, 2003 11:30 pm

Okay, I downloaded the "All Files - Full" install file and installed it. I'm not sure if I should have, but then I followed that with installing the 0.91 update patch.

I still get harsh sounds when I try to play an Alto along with a Tenor, in unison. Just to make sure we are trying to judge the same thing, I made a couple short clips of what I'm hearing. I set the random variations to "0", turned down the LFO settings, and turned off the effects--to make this test more accurate.

The first clip is a series of "B" notes at different octaves. You'll notice that some of them are more in tune than others. Some are still pretty far out. I'm not sure if I got the newly improved samples installed. Maybe I messed up the install somehow?

http://webpages.charter.net/glenweb/Tubi/4th_Note.mp3

The second clip is simply alternating between a "C" and a "D" note. Sometimes the "D" is played legato, and sometimes not. When it's played legato, I find the tuning acceptable. When the "D" is the initial note (non-legato) I find it rather shrill for my tastes. I assume that a different sample is being triggered when the note is played legato, as opposed to when it is being played staccato?

http://webpages.charter.net/glenweb/Tub ... nitial.mp3

Am I being too picky? If so, someone please let me know. I realize that a sax isn't known for having perfect tuning, but is this amount of imperfection normal?

thanks,
McLilith

WilliamK · Post by **WilliamK** » Tue Jul 29, 2003 12:05 am

Maybe I'm wrong, but I just tested here again and all Saxs are in perfect tune. Make sure you did really install the new files.

Regards, WilliamK

WilliamK · Post by **WilliamK** » Tue Jul 29, 2003 1:45 am

ANOTHER UPDATE 0.92

Here we fixed an internal memory problem, that was crashing ReWire.dll (OrionPro, Tracktion, Chainer)

Regards, WilliamK and Luigi Felici (aka Liqih)

McLilith · Post by **McLilith** » Tue Jul 29, 2003 4:55 am

Maybe I'm wrong, but I just tested here again and all Saxs are in perfect tune. Make sure you did really install the new files.

Yep, I have the new files. Just to make sure, I removed Tubi completely, including my install files. I even removed the directories, and emptied my internet cache. I then downloaded a fresh copy of Tubi your site. After I did that, you updated from 0.90 to 0.91, and I downloaded everything again.

Eventually, I re-installed Tubi and verified the version number:

Then I made the following recording of an Alto and a Tenor sax simultaneously playing the same MIDI notes, from the same MIDI stream. The recording is simply G-A-B played in sequence:

http://webpages.charter.net/glenweb/Tub ... e_Note.mp3

The B note at the end sounds very shrill to me, so I made two more recordings--this time of the Alto and the Tenor playing the same pitches alone:

The Alto:
http://webpages.charter.net/glenweb/Tub ... e_Alto.mp3

The Tenor:
http://webpages.charter.net/glenweb/Tub ... _Tenor.mp3

This shows that there was no strange waver in either of the sax voices. The beating that can be heard in the first recording is caused by a pitch mismatch between the voices.

So I have the latest version, and it is properly installed. I personally think the tuning should be closer than what it currently is. After all, if you have assorted temperaments to pick from, how can they be considered accurate if your basic samples are off-pitch?

For your reference, I checked the pitches of those B notes, and one sample was off 11 cents, and the other was off 7 cents--in the opposite direction. The beating between them is a frequency of over 5 Hz.

At this point in time, Tubi is certainly a fine Sax synth for solo parts. The tuning issues possibly make it even more realistic in that role, but I probably wouldn't rely too much on multiple Tubis playing together--at least not the way they are currently tuned.

take care,
McLilith

liqih · Post by **liqih** » Tue Jul 29, 2003 4:11 pm

I personally think the tuning should be closer than what it currently is. After all, if you have assorted temperaments to pick from, how can they be considered accurate if your basic samples are off-pitch?

absolutely correct McLilith
we have to take care of that, just that Sax samples are very hard to make tuned loops, since if loop is too short it sounds electronic,
if the loop is long it includes the irregular pitch oscillation from the player who recorded the sample,

we are working on that, thanks for the help, keep it up, <grin>

btw I loved the tune you sent, your one, great bass , what was used?

McLilith · Post by **McLilith** » Tue Jul 29, 2003 5:46 pm

btw I loved the tune you sent, your one, great bass , what was used?

I assume you mean the "McLilith TubiLeSax Demo #1" ?

I don't have the final song file for that one, so I'll have to rely on memory. I know that I used two bass instruments that were mostly played in unison, except for a few flourishes here or there. One bass is a synthetic growling synth with portamento, and the other was a standard bass guitar sound-- a smooth, deep bass without slapping or picking noises. As for instruments, I probably used Triangle II for the synthetic bass, and probably SampleTank FREE for the guitar-style bass. Drums were possibly handled by the DrumWorks DS8, another free VSTi. (Can you tell I don't have any money to spend on software?)

This was a hastily reworked snippet of something that I had done earlier, before the benefit of VSTi's. I wasn't sure about this version, except that it did seem to highlight Tubi fairly well. I'm glad you liked it!

As for the tuning issue in Tubi, I have an idea. If the samples are so hard to re-tune directly, then why not use two tuning scales?

One scale would be used to simply make the samples closer to a standard pitch. That scale file would probably be hidden and not user-accessible. Let's call that one a Pitch Correction scale.

The other scale file would be the one that users are now seeing, and would be used to further modify the tuning of the sample set, to create alternate temperaments like Dorian, Lydian, Hammond organ, Highland Bagpipe, etc. Let's call that one the Temperament Scale.

With what I have in mind, each sample set would need it's own pitch correction scale, the Alto would need a different pitch correction scale than the Tenor, for example. Of course, all samples would share a common set of temperament scales.

Might this approach be easier? Especially consider if it would be easier to work this way when you eventually record new samples in the future, to add let's say a baritone or a soprano sax to Tubi. You might take what you have learned with Tubi and create a VSTi that handles all the different wind instruments: brass, reed, woodwind, whatever. (Dare we include wind chimes?)

So, would this possibly work for you? I know it means some re-coding as opposed to re-tuning, but if you are going to continue making sample-based instruments in the future, it might save some time in the long run. Besides, you have already written code that retunes the samples, based on one scale. Hopefully, adding a second scale wouldn't be breaking too much new ground.

take care,
McLilith

WilliamK · Post by **WilliamK** » Tue Jul 29, 2003 7:34 pm

Good news, thanxs to a lot of work, we did find all tuning problems. We are also using now a Digital tuner instead of comparsion with a sine waveform, is much better.

Expect a new release with perfect tuned waveforms still today.

Regards, WilliamK and Luigi Felici (aka Liqih)

NUSofting

KVR Audio

TubiLeSax previous tests