Device Details

Device Overview

Name/Version: MaStudio 1.02
Author: nagra  
Description: MaStudio is a midi effect based on the Magenta Music framework.

MaStudio continuously records incoming midi notes and sends them in intervals to the Magenta Framework for processing. The result is then output as Midi.


MaStudio includes the following models:

continue - Interpolates a melody or a drum groove

drumify - Creates a drum groove from a melody or tap

groove - Humanize a drum sequence


These modes should be familiar from MagentaStudio.


In contrast to MagentaStudio (https://www.ableton.com/en/blog/magenta-studio-free-ai-tools-ableton-live/) MaStudio runs directly as Java-Script in M4L Node.js.
For this I changed the Magenta Music and the Tensorflow framework so that local model files can be loaded directly. So it is no longer necessary to use a client/server environment.
Furthermore MaStudio uses the C++ backend of libtensorflow for maximum processing speed.

I compiled libtensorlow without AVX on a MacPro 5.1 running MacOS 10.13, so it should also run on a Mac m1 running Rosetta.

Since I don't have a Mac m1 or a Windows system and don't release untested versions, MaStudio is currently only available for Mac Intel.


Build with Max 8.3.3 and tested in Ableton Live 11.2.10.



If anyone is interested in a Mac m1 or Windows version and is willing to test a pre-release version on their system, please send me a short request.
 

Device Details

Tags effect, other
Live Version Used: 11.2.10
Max Version Used: 8.3.3
Date Added: Mar 16 2023 18:07:52
Date Last Updated: Not updated yet
Device Type: midi_device
Download URL (report link) https://nagra.gumroad.com/l/ovizt
License (more info): None


Comments

this thing rocks, thanks!!

Hi
I'm interested in testing a windows version

@iain

Unfortunately, I have bad news. After I read that you want to test a version for Windows, I made a test with Windows 8 under VMware. Unfortunately it does not work. I get a DLL loading error for libtensorflow. But according to the documentation it should work from Windows 7 upwards.

A short search on the web gives many hits with problems under Windows. Suggested solution is to build a local version using the developer tools. But I don't have any dev tools for Windows.

I'll look at this again when I have more time.

@iain

I found something that might help.

Nao Tokui has a Windows version of his M4L.RhythmVAE on GitHub. It also uses libtensorflow. If this M4D device works for you I can use the lib from it.

M4L.RhythmVAE can be found at:

https://github.com/naotokui/RhythmVAE_M4L#m4lrhythmvae

On the right there is a releases link and the third entry leads to the Windows version.

I would like to try out an M1 processor version if you have it! The intel works fine, but Im trying to port it to my laptop that is M1. Thanks

I don't have a version for M1, sorry.
The only tensorflow lib for M1 I have is version 2 and AFAIK magenta requires version 1. Unfortunately, without access to an M1 machine and considering all lib dependencies, I can't do anything useful.

Login to comment on this device.

[ browse device library ]