Noise Canvas

It’s been decades since I bought Metasynth, which was at the time way out there in its own mind space. Paint sounds! Carve out the bad bits and tidy up the good bits! Make weird spectral beats that sound like science fiction! It could do a lot of things that required much skill to get more than a burbling mess.

After I’d sunk a fair bit into upgrades that didn’t make it easier I thought to drop it and wait until a Windows version came about.
Which never happened. I don’t think it will ever happen.

There is Virtual ANS which is free and does the basics of sound creation. Pretty good, but not Metasynth. And then there was Izotope IRIS which quite frankly screwed the pooch by supplying paint tools that had no relationship to musical or sound design. Like cutting hair with a chainsaw. But thanks for all the samples that came with it.

And then there is Steinberg’s SpectraLayers Pro which I really hoped would be like Metasynth. But it’s not at all for making sounds, only removing them. If you are making a film and the sound is dodgy in a scene you can carve out the frequencies. But then again there’s free tools around that can do that.

So here, after some considerable time, is Noise Canvas. And I’m well and truly chuffed. After following the built in tutorial I loaded up a drum beat by the Buzzcocks and carved it into a police siren. I cleaned up an old cassette and got rid of a synch tone (but dedicated noise reduction was better). I rubber stamped a sound all over the sound spectrum. I took the reverb off a camera recording. It’s fun. And if you stuff it up you can undo each step as needed.

I understand that the interface is grey to make sure you see subtle stereo effects. But a little of colour pizzazz in music creation modes might be useful.

This doesn’t do everything musical that the current Metasynth can offer. But it does the vast majority of it: cleaning up bad noises, adding weirdness where wanted, locking to BPM… it’s the tool that I had in mind decades ago. It also has most of the ability of SpectraLayers, but not specific subroutines for e.g. splitting vocals from music. But then free Ultimate Vocal Remover can do that.

Pearl Clutching and AI Qualms

So now we come to a general note about AI and creativity. Noise Canvas is co-authored by Claude – for some people that’s the end of story – the software is tainted by robots stealing the work of others.

I understand that I may have a tool in mind but not the expertise to describe its abilities and functions in a coherent manner. When someone clearly describes a tool – physical or software – they are performing a task essential to the outcome. You will recognise this when you try to use a badly written manual – the tool is hard or impossible to use because it has no proper documentation. I see the author Rob Clouth as the person that clearly documented the tool, such that it becomes useful. I see Claude as the mechanism by which the specification is delivered to the end user. I’m the end user.

I also have to point out that U&I Software has been unable to make a Windows version in decades. I am sorry that they have had trouble. But somebody has done it in months, not decades.

Sensible Review here.