Tech Preview: Hive Wavetables

[pdxindy]

so when is the first beta????????? hehehehehe

Mid of March projected for internal beta team. Gotta fix a lot of stuff before that, but need some normal days for that without NI sales messing with my brain.

cool!

(that was the plugin spirits teaching you a lesson for starting to do sales! )

[Urs]

(that was the plugin spirits teaching you a lesson for starting to do sales! )

[cytospur]

I had an idea for making E-Mu style z-plane filters using envelopes to define the transfer functions and multitables. Here's the undocumented uhm script. I think it works, I've used mostly random numbers here and intend to spend some time working out what works best, but as it is sounds pretty good to me:

Code: Select all

Info "Z-plane filters\nBy Mark Holt\n"
NumFrames=256
Seed=1111111
Wave target=aux1 "2*phase*pi"
Wave target=aux2 "sin(1*aux1+3*sin(3*aux1)/(1-0.9*cos(1*aux1)))+0.5*bandpass(rands,0.15,0.5)"
Wave start=0 end=63 "aux2"
Envelope curve=logarithmic L0=0 T1=0.02 L1=2 T2=0.1 L2=2 T3=0.13 L3=0.25 T4=0.5 L4=0
Spectrum "x*env((phase)^(table+1))^0.5"
Wave start=64 end=127 "aux2"
Envelope curve=exponential L0=0 T1=0.01 L1=12 T2=0.01 L2=-1 T3=0.096 L3=2 T4=0.2 L4=-1 T5=0.7 L5=0
Spectrum "x*env((phase)^(1-table))"
Wave start=128 end=191 "aux2"
Envelope curve=linear L0=0 T1=0 L1=-0.5 T2=0.27 L2=2 T3=0.08 L3=1 T4=0.4 L4=2 T5=0.25 L5=0
Spectrum "x*env((phase)^(table+1))^3"
Wave start=192 end=255 "aux2"
Envelope curve=logarithmic L0=0 T1=0 L1=0.1 T2=0.54 L2=-2 T3=0.1 L3=1 T4=0.2 L4=-2 T5=0.01 L5=0.25
Spectrum "x*env((phase)^(1-table))"
Spectrum lowest=0 highest=0 "0"

Normalize Metric=peak Base=Each

Set multitable to 4 and apply separate LFOs to wavetable position and multitable position (both set to 50% in the hexagon) to get some nice morphing.

It's the same base waveform used throughout (based on the feedback FM approach that I posted in Urs' FM tutorial thread).

[pdxindy]

I had an idea for making E-Mu style z-plane filters using envelopes to define the transfer functions and multitables.

Cool... will try it out!

I have a request since you seem pretty good with the uhm scripting... I have been making some vocal wavetables in Icarus and importing them into Hive. In Icarus there is the morph function and one of its options is formant. Using the formant morphing modulated with keyscaling, it gives a wider range of notes that sound legible and without the chipmunk effect.

Would it be possible to make an uhm script that took such a wavetable and acted as the formant morph mode does in Icarus?

[Urs]

Would it be possible to make an uhm script that took such a wavetable and acted as the formant morph mode does in Icarus?

No, it needs to be done on playback level. It can be faked using multi-tables, at the expense of resolution.

I already had the option (but hadn't compared to othe implementations), but I didn't like it. Maybe I need to check out Rapid and others to see what it could be like, and then reimplement it. Not anything I can do immediately though.

[drzhnn]

In wavetables you have to merge pitch and formant information into single waveform. So when you change the pitch of such waveform the embedded formant frequencies move accordingly. That's what creates that chipmunk effect. You can avoid this by using Hive's peaking filters instead. I found them very good for that particular purpose.

[StrikerTheFoxy]

I've tried the demo of Hive 1.2, and I'm wondering if it is possible to convert the .uhm files to .wav. so I could use them in Icarus and Rapid.

[drzhnn]

I've tried the demo of Hive 1.2, and I'm wondering if it is possible to convert the .uhm files to .wav. so I could use them in Icarus and Rapid.

Yes. You'll need to add a new line at the end of the uhm script that you want to convert:

Code: Select all

Export "some_filename.wav"

When you reload the updated script in Hive, a new some_filename.wav file will be created in the same folder.

[StrikerTheFoxy]

I've tried the demo of Hive 1.2, and I'm wondering if it is possible to convert the .uhm files to .wav. so I could use them in Icarus and Rapid.

Yes. You'll need to add a new line at the end of the uhm script that you want to convert:

Code: Select all

Export "some_filename.wav"

When you reload the updated script in Hive, a new some_filename.wav file will be created in the same folder.

Thanks! It worked, but it isn't working on Big Bells.uhm, each time it loads i keep getting "Error in line 1; unknown command." Could you tell me what the issue is?

[drzhnn]

Thanks! It worked, but it isn't working on Big Bells.uhm, each time it loads i keep getting "Error in line 1; unknown command." Could you tell me what the issue is?

Hm. Worked as expected here. I added Export "Big Bells.wav" line to the end of Big Bells.uhm, saved it and I got my wav. Maybe you're not using the latest version of Hive? IIRK there was a bug with single quote symbols in previous builds. If that's the case, try to comment out or delete the whole Info line.

[Urs]

"Error in line 1; unknown command." Could you tell me what the issue is?

We added Export in the final release version of 1.2. You're probably still running the preview version?

[StrikerTheFoxy]

It worked with the other tables in the preview version, and I downloaded the latest version.

[cytospur]

I've been working on a simple GUI interface for UHM scripts. I've called it UHMate

The idea is to load up a starting text document with variables in (in the example that I've uploaded there are 4 6 variables), use the GUI to change the variables and then export these as a UHM script.

In the example that I've uploaded I have included a simple feedback FMish script with the number of frames to set in the table as a fifth seventh variable.

If you want to make your own script with four variables then the variable names are:

f1, f2, md, fb

EDIT: The program has been updated to include 6 total variables (plus NumFrames).

If you want to make your own script with six variables then the variable names are:

f0, f1, f2, pow, md, fb

You should make sure that none of these terms appear elsewhere in your script.

EDIT: In this particular example, which is an alternative approach to feedback FM (where the feedback frequency can be different to the carrier frequency), the variables ModN F and ModD F (feedback modulation frequencies), Car F (carrier frequency), P.D. (phase distortion) and Feed (feedback depth) are fixed in the script, whereas Depth (modulation depth) sets the maximum value of a range from 0 and interpolated using table.

This is the base script:

Code: Select all

NumFrames=frames
Wave target=aux1 "2*pi*(phase^pow)"

Wave "sin(f0*aux1+md*table*sin(f1*aux1)/(1-fb*cos(f2*aux1)))

I'll try to find some time to make something more generic at some point.

The main point of this is as a proof of concept. It was quickly knocked up in Flowstone, so is, therefore, Windows only. It's a standalone exe and will run on 32- and 64-bit systems. I'm sure that someone with better skills will be able to do something much better than me! Just remember where you saw the idea first

Here is the link:

https://charlesdickens.neocities.org/UHMate.zip

The zip file contains the program, the starting text file and the saved uhm script.

Here is what it looks like:

[Teksonik]

Thanks but UHMate generated an odd warning in Bitdefender.......

[cytospur]

Weird. I'll test on another computer to see if I get the same.