Updated Matilde Tracker

[george]

I spent a few hours updating the Matilde Tracker machine (32-bit only) and improving the sample interpolation. This update brings significantly better audio quality with SINC interpolation that includes smart anti-aliasing and AVX2 optimizations. You'll get professional-grade sample playback in real-time (SINC 64) and even higher quality for your final renders (SINC 256). It's fully compatible with your existing projects. Readme.txt included with more details. Enjoy.

[magmavander]

Woaw!!! Thank you

[thepedal]

This is legendary. Thanks!

[Buzztler]

I still have to check it out, but pretty cool idea. Someone before criticized Matilde's Soundquality as very worse on his website ... A good thing the source of MT is still available ... on github I think ... Dankeschön!
(Edit: If it works, don't forget to upload it on joachips website, the robotplanet thingy ...)

[IXix]

Nice.

[george]

The OP zip now contains the correct name, Matilde Tracker.dll. The previous Visual Studio project had an incorrect output filename.

[Joachip]

https://buzz.robotplanet.dk/machineinfo.php?id=7996
Great stuff! Uploaded. But I have a few comments...

The recent versions are called "Matilde Tracker2.dll" (versions 1.7 and up). I wonder if we should stick to this? ... or is it called "Matilde Tracker.dll" in people's Buzz installs?

Also, It seems to use linear resampling for the first few 300-350 samples or so. Is this expected? You can hear it as a slight "tschk" sound at the beginning of each note in the attached file (which is set to half sample rate).

Also, why is this inverted? "Values: 0 = Enabled (default), 1 = Disabled"
Everywhere else "0" means disabled and "1" means enabled.

The attached audio demo is the same thing played four times with filters 0, 1, 2, 3

[thepedal]

Yes. Apologies for being too lazy to check but does this just follow on from "Matilde Tracker v1.10 rc1 for Buzz" which has native multi I/O support?

I seem to have

Matilde Tracker.dll:
Matilde Tracker v1.10 rc1 for Buzz

Matilde Tracker2.dll
Matilde Tracker v1.9 rc8 for Buzz

Matilde Tracker3.dll
Matilde Tracker v1.10 rc1 for Buzz ( think I've only used this one in the last decade and not the others)

[Buzztler]

Just if someone is interested who tested "matilde tracker (1, 2, or 3, we all don't know, - hope he knows), and some other gear, here is an adress https:// www. simonv. com/tutorials/quality.php (without the blanks of course) ...

[george]

The recent versions are called "Matilde Tracker2.dll" (versions 1.7 and up). I wonder if we should stick to this? ... or is it called "Matilde Tracker.dll" in people's Buzz installs?

Output binary in the project was called Matilde Tracker.dll for the 1.10rc build. I kept this name, but I can change it if another name is required.

Also, It seems to use linear resampling for the first few 300-350 samples or so. Is this expected? You can hear it as a slight "tschk" sound at the beginning of each note in the attached file (which is set to half sample rate).

Well spotted, it was a bug. It's fixed in 1.12, posted in OP.

Also, why is this inverted? "Values: 0 = Enabled (default), 1 = Disabled"

Yeah, it makes sense to have 1 enabled. I reverted that and set it as the default. Other bugs have been fixed, and there's a new SINC 32 mode that complements the default SINC 64 mode. SINC 32 is half as CPU-intensive for real-time tasks with very similar quality. For further optimization, if the sample is played at the root note, no interpolation will be applied.

[george]

Fixed bugs with long wave filenames handling, interpolator cleanup, and SINC playback issues with loop points. It should be ready for production use.

[skizzo]

Osom, man!

[Joachip]

Awesome!
Uploaded v1.12 to the buzz site.

[thepedal]

SINC 32 filter mode seems to really pop and click for me. Running on a recent laptop with an Intel Core Ultra 7 165U and a decent rme asio soundcard. Is it just way more intensive than linear/spline and I should just stick with one of those? At the moment I'm just testing trying to trigger mono drum hits which are 'RIFF (little-endian) data, WAVE audio, Microsoft PCM, 24 bit, stereo 48000 Hz'. Card is running at 48000 Hz with the asio at 2048 samples buffer as I don't really care too much about latency.

[thepedal]

Ok, so it's more esoteric than I thought. If I create a new empty .bmx with an Mtrack 1.12 (about page still says 1.11 btw, no biggie) and have that controlled by a peer chord so it's playing kick drum samples through a range of notes then 'SINC 32' filter mode '3' and 'Offline SINC 512 Bounce' '1' doesn't produce any clicks or pops. Where I have the problem is in tunes that were created using 'v1.10 rc1' which are now forced to load 1.11/1.12.

If I take the tune, remove all the Mtrackers, save it and then on re-opening add in just 1 Mtrack 1.12 and set it to start playing these kicks again I still see/hear the same thing - sporadic loud clicks and pops. Watching the buzz cpu monitor that Mtracker seems to spike to 10-12% cpu sometimes when it's triggering. There's still plenty of overhead though, Windows resource monitor is not reflecting these spikes.

[george]

Yes, I noticed some issues when loading old versions and added some sanitized init to fix that.

Please try the new 1.13 build and report back. Thanks!

[thepedal]

Cheers I'll give that a bash tonight.

[skizzo]

Thanks a lot - will try out, too!

[thepedal]

Afraid it's only very slightly better, if at all. With filter mode 3 enabled on 3 instances of Mtrack 1.13 (about page update thanks ) I see in Buzz cpu monitor they all spike into the 10s of %s and the pops and clicks are fairly constant if not regular in any way if that makes sense. Will quickly try deleting and re-adding those machines but I suspect it won't help...

[thepedal]

Yeah still not fixed sadly. Sent you a .bmx example if you want to see if you can replicate. Thanks for your work so far, it's amazing were still getting improvements to Buzz stuff even after all this time