You've got questions? We've got answers!

We're a small company, so up until now it's felt a bit silly publishing "Frequently Asked Questions" because, let's be real, when you have less than 100 clients, not many questions are frequent. However, there are certain questions we receive often enough and folks encourage us to write the answers up in a blog post.

This is that post! Without further ado, a curated list of "infrequently-but-sincerely asked questions" that we've deemed blog-worthy.

What's the difference turning "Eager Mids" on or off?

In "Eager Mids" mode, the plugin acts more aggressively on the range from 100-1000Hz. You can see it in the visualizer and hear it in the resulting audio. Most folks like this setting better, which is why it is on by default.

The "Eager Mids" option makes use of a larger "receptive field" for computing the changes to be applied. Simply put, the "receptive field" is the amount of audio context that the plugin has to ingest in order to produce output. Non-"Eager Mids" consumes a context of around 100ms, while "Eager Mids" consumes 300ms. That increase causes a slight bump in the required computing power by the plugin. We're continuously doing R&D in order to provide more options for the plugin and enhance its performance in future updates!

How frequently will Deprox be updated?

We plan on updating Deprox as we obtain more data for training the machine learning network powering the plugin at a bimonthly cadence; these updates are focused on refining the overall sound and feel of the plugin. Research and development is continuously being done in aspects as such ML engineering, plugin performance, and additional functionality; we aim to release features as they stabilize to ensure a smoother experience when transitioning between different versions of the plugins.

If you'd like to know where specifically our attention is going these days, check out the RTNeural project. We have a whole slew of high-quality trained models that we can't deploy yet because they aren't fast enough on most machines. This project has the potential to solve that problem.

What is a pre-release version for Windows?

Pre-release means that internally, we're building the plugin binaries manually compared to going through our established release flow set up through a continuous integration environment. At our team of two in Meeshkan Plugins, a good chunk of plugin development is done on Mac machines; as such, we are much more familiar with its development environment and ecosystem compared to Windows'. If you require a VST3 plugin binary for Windows, do get in touch through our Discord server (https://discord.com/invite/JDXTsU9Bn3). We plan to automate the release flow for Windows in the future such that this intermediary step of requesting a build isn't needed.

What's next?

When we set out to build audio plugins in Meeshkan, we made a list of "really hard things" that we vowed to check off one-by-one. We aim to make use of novel approaches in the AI-based signal processing domain to deliver products that we feel check off things in this list.

0 comments

Everything We Learned While Building an Audio Plugin Using Machine Learning

Hello Deprox!

See all posts from Meeshkan