DSP Robotics Support

Hi gang,
here is my latest chorus effect. The goal was to create something that sounds like a real chorus - a number of musicians playing the same part simultaneously. Well, I failed. However, I think the result is still interesting if you just want to enhance an otherwise dull instrument, add stereo width, or - at extreme settings - generate some crazy noise. You tell me.

That’s a great sounding chorus effect Martin and, especially considering the amount of delays, the CPU is really low.

I’ve always been attracted to the idea of turning one instrument or voice into many slightly different ones, but I found the brain is a hard organ to fool in this respect (and many others). We can hear the identical nature of the individual delayed sounds, even with the slight frequency variations and panning.

I suspect (but I don’t know) that we need to consider the phases of all the partials making up the sound, and scramble them somewhat for each instance. That would involve a FFT-phase scramble-iFFT system which is beyond me, but maybe not beyond you. In addition it might be of benefit to vary the amplitudes of at least some of the partials. The problem then, I guess, would be the time taken to perform the processing. Even a small delay to get the required effect would add latency, so it may have to be an offline process.

Another aspect to consider is the spacial effect. If you imagine a choir, not all voices originate from the same point, so there will be subtle time differences between the arrival of the wave fronts and between the listeners ears, quite apart from differing early reflections and subsequent room reverberations.

It’s a fascinating topic for sure.

As always, many thanks for sharing your amazing work Martin.

Cheers

Spogg

Spogg wrote:I’ve always been attracted to the idea of turning one instrument or voice into many slightly different ones, but I found the brain is a hard organ to fool in this respect (and many others). We can hear the identical nature of the individual delayed sounds, even with the slight frequency variations and panning.

Yes, exactly! When two instruments play the same melody, each note will sound different, and the difference will be different for the next note. They may be in tune on one note and out of tune on the next. They may hit one note in sync and be off by 50 miliseconds on the next. Timbre will be different, too.

69 voices - reminds me of Steve Martin's Googlephonics routine; 'cause ya know with only 67 voices it just sounded like sh*t!.

69 voices? The more the better right? i bet it will sound great

martinvicanek wrote:Well, I failed. However, I think the result is still interesting...

"Honour thy error as a hidden intention", as Brian Eno's Oblique Strategies might advise! (and here's an on-line random Oblique Strategy picker - it works wonders for my pathological procrastination! )

It certainly is a very interesting sound though, and I like that you've given the controls a wide enough range to get some really chaotic sounds out of it. At more subtle settings it makes a good early-reflections simulator to add a bit of space without a noticeable reverb tail, which I'm rather taken with on drums that are a bit too dry sounding.

Wow Martin,
Really top notch!
I really like the sound of this chorus, it does so much more than what you would expect from just looking at it, but when you start to use it, it really shows it's stuff.

Again great work!

Later then, BobF.....

Thanks guys.

When I was a boy my grandma used to take me to church. I remember the fascinating effect of the crowd repeating after the priest, this diffuse murmour caused by the spread of timing and the abundant reverbration.

Another memory I have is when a song was sung, the onsets were notoriously late (and diffuse). Like if there was a huge inertia to overcome for the sound to develop. Probably people were just hesitant about the words or something.

Anyway, I always wanted to create such effects, and I think Chorus 69 is about half way there. Wonder if I should add a preset manager, Probably should.

^ A slight 'churchy' connection here, too - using some of the more extreme settings on drum loops made me think of the Boy's Brigade marching band I was in as a kid. You can get both the sound of dozens of drummers who can't play in time with each other AND all the slap-back echoes of the surrounding streets!

I think subtle timing variations for the start and end of discrete sound events is one important aspect of chorus simulation. But how would you do that with DSP? The DSP would have to change these values on a per-event basis, so it would need to look into the future. Just randomly modulating delay times surely wouldn’t cut it.

One approach might be to only update timing and other variations during short periods of silence in the incoming stream. I suspect that would only be good for certain types of input though, like totally dry vocals or percussion. Otherwise it would likely be a click-fest.

I plan to experiment, at some time, with the idea of making a synth where all the relevant parameters are in place for a more convincing chorus or ensemble effect. The advantage of this would be that any processing could be on a polyphonic per-note basis. The system I have in mind would feature random start and end timing offsets, pitch errors and partial phases (pre-processed offline) and adjustable static panning. Each voice would be applied to independent mono echo/reverb modules to simulate slightly different early and late reflections.

I think I’m gonna need a bigger CPU.

We shall see (hear) what happens…

Cheers

Spogg