A recent update to Cockos' Reaper – version 5.97 – could have gone unnoticed this month. In this update, the community-friendly DAW just added Melodyne communication via ARA 2.0 to its long list of features, making it the sixth major piece of audio host software [by my estimation] to integrate the technology, alongside Studio One, Logic, Cubase, Waveform and Cakewalk.
Actually, the number of major developers taking notice of the ARA standard is now bigger than the number of developers who create the tools that use it, but we'll get into that later.
Despite the uptake, you'd be forgiven for not knowing what ARA is, and forgiven moreso for not knowing how it works and what its advantages are. It's time for some prosthelytizing…
What it is
ARA stands for Audio Random Access, and was originally conceived and built by Celemony, creators of Melodyne, and PreSonus, creators of Studio One. Back in 2012, the basic upshot was this: if you had Melodyne installed, you could call it up and operate it within Studio One, editing pitch using Celemony's software within PreSonus's usual DAW timeline, without needing to actually launch Melodyne and shuttle audio back and forth.
That's how it seems from the eyes of the user – and mightily impressive it is too – but when you see it from the eyes of the developer, you get a little more insight into the potential behind ARA.
How it Works
A standard plugin (VST, AU, et al), is fed audio data by the DAW when playback starts. While latency can be introduced behind the scenes to allow things like lookahead, the plugin only receives short buffers of the audio played by the playhead – you can look at it as a real-time stream, although in reality it's a bunch of short bursts that are later converted back into a real-time stream.
ARA is more sophisticated. Instead, it receives comprehensive information from the DAW about all the audio that's been placed in the entire project: what it is, where it is, how it's trimmed – all to one-sample accuracy. This means that when you're processing audio on one channel, not only does the processing software know what's coming up much later in the timeline, it also knows what's going on on other channels.
You can think of the difference in this way: Imagine loading up a VST plugin designed to normalize audio. Normalizing is a process that detects the highest peak in an audio signal, so the plugin would have to be fed the entire audio signal in order to determine what the highest peak is, before being able to actually do anything to it. This would require a 'detection mode', a subsequent 'processing mode', and inevitably, lots of clicking and time from the user. It's not impossible, but it's certainly impractical.
An ARA processor, on the other hand, would have access to all the information about the audio on the track, and would instantly be able to define the highest peak in the signal and the processing needed to normalize the content. It would also be able to take other tracks into account while doing it. Imagine the power of ARA for masking detection, an analogue-style replacement for your virtual mixing console, altogether new mixing paradigms, audio repair, or anything else.
The ARA Landscape
It's up to ARA 'client' developers how much they make use of ARA's capabilities, and it's up to DAW developers just how much control they'll give the user over ARA-compatible software within the DAW. Just because Studio One 4, for example, can give you access to Melodyne within the timeline from a quick key command, it doesn't mean another DAW won't put your ARA software into a floating window, or even potentially hide it behind abstracted controls.
So let's get back into context. There are three companies currently creating ARA-compatible software, and they all happen to be in the business of pitch manipulation. Celemony, who were one responsible party in creating ARA, allow Melodyne to be used in harmony with your DAW; Synchro Arts, developers of VocAlign and Revoice Pro, also support the technology; and the latest addition is the first name in pitch software, Auto-Tune, with the Pro version of Anatres' package being compatible, too.
The problem of ARA adoption can be compared to a service like Uber, where success depends on encouraging both drivers and riders to start using the platform. ARA's uptake has been moderately paced, but the number of DAW developers who now support the platform might be reaching a critical mass. But with the number of ARA client software packages numbering only four, the platform needs a higher uptake from plugin developers if it's going to take off. This is a fairly big ask, though, with plugin developers already having many standards to meet in terms of formats, DAW compatibility and so forth. The current roster of ARA-compatible software is mostly made up of programs that need to run standalone rather than being ready-integrated with the DAW (the notable exception being Auto-Tune Pro).
So where next for ARA? I'd love to see the technology picked up by players such as iZotope, whose RX and Neutron packages seem like the perfect match for the technology. Indeed, anyone into developing clinical audio tools should be a natural fit, but it's the creative uses that I'm looking forward to being surprised by. Music itself has always been shaped by the technologies used to make it, and the technologies themselves usually appear as ways of making complicated things easier. With a larger ARA ecosystem in place, it'll be up to musicians to push the technology to its limits.