Datamind Audio have announced the launch of a new AI-powered plug-in that’s capable of turning any sound into a playable instrument. Described as an “audio mosaicking tool”, Concatenator can be used to create complex new sounds and textures based on any audio file or input source.
The innovative new instrument relies on a new form of synthesis which has been developed by Datamind Audio. Named Concatenative Sound Synthesis, the approach is said to be similar in concept to granular synthesis, but with the grain selection, amount and size parameters influenced by an audio input signal rather than being chosen at random. It’s possible to feed just about any audio into Concatenator — it’ll happily accept input signals from a mic or instrument, as well as existing samples — and the instrument will respond by analysing the signal before ‘reconstructing’ it using fragments of other sounds from the user’s sample collection.
Created for sound designers and musicians alike, the plug-in offers a quick and easy way to create complex, layered sounds, and is capable of generating everything from foley and soundscapes to synth-like rhythms — all with the user’s existing sample collection. The company say that at present, Concatenator incurs a significant amount of latency (around 900ms), and so is not currently capable of being played live in real time. The limitation is due to the intensity of the underlying processing, and while there are ways to speed it up, Datamind Audio say that the solutions currently come at the cost of reduced audio quality.
Concatenator’s generative AI models create new sounds by ‘imitating’ the timbres they’ve been trained on, and rather than equip the instrument with a large, all-in-one model, the company instead offer a range of Artist Brains. Each Artist Brain has been created using a database of sounds supplied by artists that have partnered with Datamind Audio, and so offer their own unique take on the processing that’s applied to the input signals. Of course, there’s plenty of controversy surrounding the use of AI and machine learning in music, and company are keen to point out the following: “Artist Brains are not capable of plagiarising artists. They do not make music for you, and they only vaguely resemble the artists the models are trained on. You can not slap the Concatenator on the master channel and hope to pass yourself off as the artist that the model was trained on.”
Key Features
- Machine Learning: Modulate how the audio-matching algorithm ‘weighs-in’ the spectral and amplitude components of the input signal.
- Lightning Fast Load Times: The only limitation on how many samples can be loaded into Concatenator is how much RAM is available.
- Flexible Modulation: Five LFOs, five envelope followers, five side-chain-enabled MIDI envelopes.
- Total Control: Control sample repetition with Variation, re-scatter the probabilities for chosen slices with Particle Reset, and then modulate how likely the sample is to hold once selected using Stickiness.
- Dynamic Presets: Presets can be used with any sample combination, while additional Corpus Presets store the currently loaded samples.
Check out the video below, where Datamind Audio’s co-founder Ben Cantil talks through Concatenator’s features and demonstrates what it’s capable of.
Compatibility
Concatenator is supported on PCs running Windows 10 or higher, and Macs running macOS 12 and above. VST and AU plug-in versions are available.
Pricing & Availability
Concatenator is available now, priced at $149. Artist Brains are priced at $33 each.
