Technical Details / FAQ
⚠️We are still in Early Access. Although we try to keep this page as relevant as possible everything here is subject to change, could be outdated, cancelled, replaced at any time. If you see any mistakes please send them to firstname.lastname@example.org ⚠️
Alex Audio Butler is a track-based VST3 plug-in. The main idea is you insert 1 plug-in on each track, based on what’s on there: Voice, Music, and one for the Master (SFX, Ambience will follow).
Since it’s a VST3 plug-in we can only know about audio that has been played back, so while you’re editing Alex Audio Butler creates an internal analysis of the audio of your project. Then generates internal breakpoints/keyframes/automation that control an internal effects chain based on that analysis.
In this version Alex Audio Butler’s main focus is volume/level/loudness/gain. So the analysis is a psycho-acoustics-based loudness measure. The effects chain is mostly gain stages combined with compression and limiting. In the future this same approach can enable smart-EQ’ing, and maybe even auto-denoising.
There are also some fixed problem preventers: low-cut to prevent rumble, limiting to prevent clipping distortion, and speech will always be made mono.
The mixing assumption is:
- You will always want all voice in the foreground, at a constant volume.
- The role of music heavily depends on the type of content, so you can choose the importance/loudness (currently, in v0.6.4, we assume 1 song, or at least songs with the same loudness per track, this will change in the future), and ducking strategy.
For more complex projects human expertise is advised.
⚠️ The auto-level algorithm is constantly in the works, based on user examples where Alex Audio Butler doesn’t sound as expected. If you encounter such a situation, please send raw audio, with description to email@example.com. The other main focus is being better and better in getting it right the first time, or at least second time, and informing the user about it. ⚠️
During exports we also keep learning, so a quick audio-only export is a good way to make sure Alex knows all. We’re researching if combining VST3 with Adobe CEP scripting helps Alex Audio Butler know more about edit changes in the sequence. That would enable us to inform the editor better about the status of the internal knowledge. In practice adding Alex AB somewhere close to the end of editing “just works” most of the time.
The plug-in instances communicate with each other, enabling ducking music when there is voice in an other track. Also the plug-ins look-ahead a bit to try to have a correct mix the first time, but the internal keyframe trick enables anticipating what’s coming on further play-throughs. This way it can do nice longer fades for ducking for example, in stead of using a side-chain compression approach which sounds very pushy in many cases, and is always a bit late.
The compression presets are possible because the first step is auto-volume, so the input level of the compressor is predictable, which removes the need to adjust the threshold (the part that makes presets for compressors useless normally).
The master instance analyzes the total loudness of the sequence (can be tracks with and without Alex AB) and creates a gain-compensation based on your export target choice. Currently we can’t guarantee LUFS correctness, because it’s impossible to know if anything changed since the last audio came in, but:
- Broadcast comes close to EBU-R128,
- Online tries to maximize loudness, without sacrificing quality
- Archive is in between: loud enough to not totally fall away between other online stuff, but less to no hard-limiting, so you can still do that later.
For perfect broadcast specs you will stil have to correct it using Izotope RX for example, but it will be close, and that way the master compression and limiting still can be correctly optimized for this loudness level.
The goals of Essential Sound and Alex Audio Butler are very similar: simplifying the predictable parts of mixing, but the workflow, and concepts are different. In AAB you assign audio-types to audio tracks instead of clips, and AAB keeps listening and updating while you edit, making a workflow with templates (even 0-click mixing) possible. Also there are of course many differences in mixing choices, and functionality. Both Essential Sound as AAB are in very active development, and it’ll be a matter of taste too.