Artificial Intelligence, AI for short, seems to be seeping more and more into the world as we know it. From healthcare to education to even real estate, the touch of AI can be felt in ever increasing regularity. It’s also now being seen more and more in the music industry and in particular plugins. A number of major industry players as well as young upstarts are harnessing its power, to enable music professionals to make and produce better sounding music. In today’s blog episode I am going to highlight 5 of the companies / plug-ins best harnessing the power of AI. So let’s dive right in.
Humtap listens to your voice and turns your hums into songs. I’m sure lots of us have had the idea to create something like this, I know I did, but Humtap actually turned intention into reality. The app records your hummed melody and chosen rhythm and applies AI algorithms to compose, arrange, perform and produce an instrumental track — all you need to do is hum your tune and select a drum track on your phone’s screen. You can even choose a musical genre, such as Depeche Mode or Metallica.
Humtap’s CEO predicts a future where anyone, even without musical training, studio equipment or financial resources will be able to make and produce music. AI will do the production and mixing for you, all you will need is a smartphone and the app.
AI has also infiltrated the Mastering arena heavily. One of the companies riding this wave is LANDR. Their AI powered mastering engine listens to your unmastered song, identifies the genre and applies the relevant mastering equalisation, multiband compression and other processing. All without human intervention. The processing is adaptive, responding to the needs of the song by continuously tweaking the EQ, compression and other processing tools throughout the track. Each time LANDR masters a track and listen to new music, the better it becomes, thanks to self-learning algorithms.
LANDR are currently mastering over 330,000 songs per month, more than all the studios in America combined. That is very impressive to say the least. It’s also gotten to a level where differentiating between a song mastered using LANDR or an actual mastering engineer is difficult. Personally for me, I still like going to a mastering session with an actual engineer and having a direct input into a session. But who’s to say this will not be possible with future versions of the service in some way and to be honest, if I were a betting man, I’d say this method of remote AI mastering is just going to grow in popularity to a point where it’s seen as the ‘normal’ way to do things.
Izotope are an American company that develops award winning audio software and plugins for mixing, mastering, restoration and more. They were one of the first companies out of the blocks to really leverage the power of AI in the plug ins. Since the 2016 release of Track Assistant in Neutron, they have been developing ‘assistive audio technology’ as they call it to help producers with their audio productions. Their stated goal is to eliminate the time-consuming audio production tasks users experience, so they can focus on being creative.
Their assistive audio technology analyses audio and provides custom presets to tailor the sound, helping you to get it to where you want.
Their assistive audio technology consists of 3 pieces:
- High level user preference
Before running any assistive tech you are asked to specify broadly the type of sound you are going for. This helps prime the assistant to get a sense of what you are going for and make the appropriate changes you’re looking to make
- Machine Learning
A machine learning algorithm characterises your audio in some task specific way, e.g. the instrument classifier in Neutron (Izotope’s mixing plugin). This information then tees up the software to make the required changes to get your audio to where you need it to be.
- Intelligent DSP
Further analysis of specific audio properties is then carried out, taking into account your user preferences to set parameters of different DSP modules like a Compression module, a Tonal Balance module etc . For example, the dynamic range of your audio is analysed and then parameter values are selected for a compressor module to affect your audio in a way that enhances it (hopefully). I say hopefully as it doesn’t always work as you’d expect in practice but more often than not the results are very good. This process is referred to as ‘Intelligent DSP’ – where DSP stands for digital signal processing, i.e. your audio signal gets affected after AI analysis and then treated with resultant AI powered recommendations.
Actually one of the competitors to LANDR is Ozone 9, Izotope’s Mastering plugin It makes use of an AI system called ‘Master Assistant’. Essentially what this does is analyse your audio and actions changes for you. You can accept or cancel these changes and also tweak them further manually. It’s very good, with lots of modes to give your music either a vintage or modern feel.
Oeksound – Soothe
This is one of my favourite plugins to use AI – Soothe by Oeksound. Originally billed as a vocal processor, it’s exceptional quality makes it superb as a processor for any audio type. It acts as a dynamic processor for mid and high frequencies.
Basically soothe is a dynamic equaliser with self adjusting frequency bands. Soothe analyses the signal in real time and adjusts the frequency based reduction based on the input signal. This saves time manually automating problematic mid and high frequencies. The reduction kicks in where needed and when needed, without affecting nearby frequency areas.
For any mix engineers or producers out there soothe really excels in the following areas:
- Sibilance issues on a vocal
- Fingerboard noises on a guitar
- Whistling type sound on cymbals etc.
- Overly bright piano and guitar tracks
This plugin does an amazing job at taming these sorts of cases and is brilliant. I’d really suggest giving it a try. It’s already a favourite plugin for mixing legends such as Greg Wells who’s mixed Timbaland, Katy Perry and Adele, amongst many others.
Another amazing plugin is Gullfoss by Soundtheory. It’s an intelligent equaliser that listens in real time to a signal and decides how to prepare the audio so that your brain can get the most information out of it. Gullfoss allows for quick and precise fixes that would require significant time and automation to fix. It really is a time saver and game changer for engineers like me, who can spend hours tweaking eq automation lanes to get them audio sounding right.
Gullfoss leverages a new patent pending equaliser technology that processes audio superbly. This equaliser can change its frequency response 300 times per second, without any audible artifacts or degrading signal quality! Currently it’s the first and only product of its kind and is consequently really making waves in the industry.
As you can see, AI is really starting to make serious headway into the music industry. Some of the services and plugins out there save so much time and effort that they are already invaluable tools in the arsenal of many a music maker. As AI gets more and more powerful will it be able to make songs? Well with companies like Juke Deck, recently acquired by Tik Tok, it already can. How well? I’ll leave that for you to explore and find out.
Personally I for one could never automate audio tracks to change 300 times per second like Gullfoss can. But at the same time I’m not sure I want to one day maybe getting fed music hits made with no human input, which could in theory could happen if the AI systems making the music become intelligent enough. Anyways, I guess at that point I’ll be getting chased down an alley by a Terminator asking why a human is its master. Whatever does happen it’s a very exciting time in the music industry with exceptional tools and services on offer that are continuing to democratise music making by allowing anyone to create it to a good level, which really is a good thing.