Last week we gave you an overview of platforms we thought could be helpful when you are thinking about collaborating online with other musicians in separate locations. Now, Red’s Innovation Manager Karim Fanous and Junior Programme Manager David Fong would like to give you a quick tour of some apps and tools that leverage artificial intelligence to help you make music.
It’s a pivotal time for artificial intelligence and music. In the past few years we’ve seen a new wave of artists embracing the idea of collaborating with artificial intelligence tools. Benoit Carré released an album using his SKYGGE moniker in 2017 called Hello World, Composed with Artificial Intelligence, in which he’d used AI tech called Flow Machines to provide chunks of audio which were then arranged to compose tracks or parts of tracks. On the other end of the spectrum Holly Herndon created an AI called Spawn to interpret singing by her ensemble and throw back music and textures to form the beds of songs.
Alongside these artist experiments, we’ve seen a wave of apps that use deep learning and other processes to generate music autonomously with minimal guidance, for example our alumnus Humtap, which we’ll tell you about later, either at a fun casual user level or more professionally at a professional score outputting level.
Some apps, like our alumnus CloudBounce, are trying to help automate certain studio processes using AI, like mastering or mixing, while at the other end of the mixing room glass smart instruments like our alumnus Vochlea are harnessing machine learning to help you make music quickly and intuitively with new sorts of instruments.
All of this makes for a very exciting landscape of digital music creation tools – the first wave of what is a very promising future AI-assisted music creation toolbox.
So, to the apps! We’ve split this into three sections: fun, short-form music generation; smart instruments; and studio automation tools. There is actually another category of platforms out there like AIVA which cater to a more professional long-form or score-based audience. We’ll save these for a separate blog.
AI-Assisted Music Creation Apps – fun and short-form
By saying ‘fun’ here we don’t want to detract from the quality of sounds and tracks that these apps can output. What we mean is that the following group of apps sit at a more accessible part of the spectrum. Some have fun user interfaces; others are designed to help you generate full tracks or motifs with minimal input. Under the hood they are using very complex algorithms and sometimes deep learning neural nets. They make very complicated music generation processes seem simple.
We’ll start with Humtap here because we are proud of it as a graduate of the Abbey Road Red incubation programme. It’s a fun and free to use app aimed at casual music makers or fans that uses a combination of techniques including deep learning with neural nets, music information retrieval, digital signal processing, musicology and sound design to generate music in real-time with a few simple user inputs. You choose a style, hum a melody, tap a beat and then Humtap generates a unique and produced track for you.
Boomy is next. It’s an online platform that, similar to Humtap, presents you with generated music, this time quicker than real-time, after you give it a few inputs. We chose rap with a Mahogany Flex filter for it to generate an initial track with. After that you can re-generate the track with a different influence or BPMs, as well as copy and paste or re-order sections of tracks. Boomy will also distribute your finished track straight to popular streaming services if you pay for its Premium or Pro versions, which also unlock other features.
Finally, JAM is an app that hands you hundreds of mix packs and thousands of loops which you can combine together in an 8-channel mixer. The app’s AI track assembling engine takes the content you have selected, in addition to your preferred FX and harmony features, and generates a track from them which you can share directly to one of many social media platforms.
AI-assisted Musical Instruments and Creative Tools
We believe that smart instruments will lead the way to a new generation of music creators, as they offer a fun and inspiring alternative to acoustic instruments and are designed to be seamlessly integrated into a studio-based environment. Here we would like to introduce a number of both physical and virtual instruments that may take you on unexpected creative journeys.
Being a drummer without a drum kit at home will feel especially hard in these times when you can’t get to your drum room or rehearsal studio. Fortunately, Freedrum has come to the rescue with intelligent drumstick sensors through which you can play a virtual drum kit wherever you are. You can build your ideal drum kit through a dedicated mobile app, as well as record MIDI patterns into your favourite sequencer through just a Bluetooth connection.
Next we have Vochlea, another graduate from our incubation programme, which has created an exciting real-time MIDI controller which learns your voice so you can directly input the musical lines and manipulate the sonic parameters of any instrument using it. The Dubler Studio Kit, costing £259, comes bundled with a microphone and a virtual MIDI instrument which work together to transform your vocal ideas into MIDI notes within your sequencer.
If you are a producer craving unlimited ideas, Orb Composer‘s Producer Suite costing €99 may be the perfect plugin suite for you. Through four dedicated plugins named Orb Chords, Orb Melody, Orb Bass and Orb Arpeggios, you can let AI generate the musical elements you need to build your next hit track in your sequencer.
Another suite of tools which may provide you with AI-powered inspiration is Musi-Co, which gives you an AI drummer as well as an AI music generator as plugins to use in your DAW of choice. If you prefer producing music on your phone or tablet, you will find the Impro AI and Song apps tempting, which allow you to control the AI-generated music through your gestures.
If you are after a more real-time jamming tool to simulate the drummer in your band, the Rhythmiq plugin will appeal to you. All you need to do is feed in an initial beat, and you get a potentially infinite amount of musical variations which you can fine-tune with respect to density, space and variation.
Talking about simulating drummers, did you notice that Logic dropped an AI drummer into its toolset recently? You can achieve natural sounding performances from any one of its 15 virtual session drummers simply by picking a genre and fine-tuning the drum pattern’s parameters, which include loudness and complexity, on an XY pad. You can then take these patterns and reproduce them on your favourite Logic virtual drum kit.
Automatic Mixing and Mastering Platforms
As many professional studios close in this period of isolation, you may well consider passing on your finished track to an AI-powered mixing and mastering platform. While not a replacement for a professional mastering service, the results we obtained through this algorithmic approach have been impressive and continually improving. We have a handful of platforms we would like to recommend here.
First, CloudBounce, an automated mastering platform. Close to our hearts, it was an early graduate of Red’s incubation programme. Its mastering engine analyses the track you upload to it and then applies a mastering algorithm that suits the track. You can also choose preset styles to master with. The cost is $9.90 per track or you can subscribe for $19.90 for unlimited usage.
We’ll tell you something else, for anyone worried about substitution: we had a deal with CloudBounce and Abbey Road where we advertised CloudBounce as an option for people who couldn’t afford the online or full mastering service. Without revealing too much detail, we found that people who use CloudBounce tended to opt for online or full mastering at a later date. So far from substituting our online and full services, it helped with a small increase in business for them.
Landr is a big player in the space, which offers a monthly subscription to its Master and Release service through your tracks are made ready to distribute. The platform’s AI algorithms successfully louden your track by learning your production style through intelligent audio analysis, adding customised audio processors, and using machine learning to set the parameters of these processors to make your track sound as good as possible. We particularly like how the intensity of a master can be switched between low, medium and high settings so that you have control over how much your track should be pushed.
If you can only afford to master your tracks for free, Popgun gives you that opportunity with Gloss, with which we obtained very good sounding results nonetheless. It runs as a simple web application to which you can upload a track and wait for the AI algorithms to work their magic on your audio material. We found the description of the changes made to the track once it was ready very useful. We’re interested to follow Popgun’s progress, as it is also working on a suite of generative music tools across different instruments as well as whole tracks. If it’s releasing mastering tools, this also means it’s using its AI engines across many different deployments from music generation to studio automation. An end-to-end or holistic AI approach and one to watch in this area.
You may also want to explore OSMix for your next radio-ready release, which delivers its automated mastering service through a downloadable desktop application and is similarly equipped with AI capabilities.
Finally, eMastered is another subscription-based mastering service that uses AI algorithms to analyse the audio content of your track, and pick out the processors it needs to enhance it so it is ready to release, all directly within your web browser. We were happy to discover the platform lets you conveniently manage all your tracks and album releases within a well-designed library.
That’s it for now. We hope you enjoy exploring. Please watch this space for a follow up on pro-level score generation and online music learning soon.