Hey,

Just thought of a great way to play chords on your mono synth. There has been a few moments since I got my Slim Phatty that I've wanted to play some chords and I have bounced down 3 or 4 different notes to see how it would sound but just found it became a time consuming trial and error process to get it right.

And then today I though it would be great if there was a plugin that transposed multiple copies of a single note and allowed you to play one fingered chords and this led me to try a little experiment:

- In Logic I sent my Slim Phatty to 3 different aux channels (and then those 3 to a single aux channel to add effects)
- Used Logic's Pitch Shifter II plugin to tune them up some semitones (0,3,7 for a minor chord)
- Voila, one fingered chords

I know it's probablt not gonna be perfect but I found this to be a great way to get a good rough idea of the chord before I bounced down the different notes properly.

download the "notelogic / voice allocator" plugin from the xhip site in the effects section.

process your polyphonic track with it. set the number of voices and the release time.

it'll output notes divided to midi channels. use your host to split the track by channels.

render/record each channel individually.

mix them.

result:
http://soundcloud.com/aciddose/eight-a-01

this one i panned my ad-x-1 to the left, and my sh-09 to the right.
http://soundcloud.com/aciddose/squiresquare

just the x-1.
http://soundcloud.com/aciddose/x1polyfour

this will turn it into a real polyphonic synthesizer. this is _exactly_ how a polyphonic synth would allocate it's voices. the only difference is because you only have one "voice" you need to render it multiple times. so it will take pretty much exactly time * voices to do. easy though. no effort involved.

I just use a VA synth (Diva) to prototype. I find that if I set things up right, the recorded chords from my synth sound pretty similar, but a bit thicker.

Does Xhip use FFT (fast fourier transform) to do this? Just bare in mind FFT usually sounds OK within one octave but start's to lose clarity after that.

what are you talking about? this has nothing to do with xhip.

fft to do what?

we're taking a midi clip - splitting the notes into channels as "voices", then playing each voice once at a time using an analog synthesizer and recording the result.

if we had 16 identical synthesizers with 16 identical presets we'd just use that instead and render everything in real-time. that would be called a polysynth.

since we usually only have one copy of our mono-synth, we need to record it in sixteen steps.

I probably should download the tool to see what it does.

But, I assumed that a tool like that would allow you to protype polyphony with FFT. FFT transposes notes up or down without changing the time. I do it all the time in MaxForLive.

I guess I don't really understand why you'd need to split a chord up in to channels. What's the point?

When I want to record a chord from a monophonic synth, I just play the notes in a sequence like an arpeggio, then take the pieces of audio and put them back in to line with each other to make a chord. It couldn't be simpler. You can hear the chords in this track. They are done with an SH-2:

http://soundcloud.com/syncretia/saturnalia

How would your plugin make this easier?

Have you got a tutorial video that shows how this would work in Ableton?

no, i ca't go making a ton of demos for every host.

it's a midi processing plugin.

set the release time and number of voices you want and make sure it matches the synth sound. it won't hurt to use a slightly longer time, you want to match or slightly exceed the time on the synth or voices will get stolen too soon.

say you have a sound with 240ms release and you want four voices, set that up. this will usually fit a typical sound with short release. it's also the default setting

now i'm not sure.

in cubase i know you can put the midi processor in the instrument channel and then you can select another channel and the plugin shows up in the list of midi inputs.

i'm not sure if you can insert it to process midi before it's recorded to a part.

in reaper you can insert the plugin and tell it to replace the midi part on a single track, and it works with your instrument plugin already inserted there.

it also lets you output the midi and record the audio in multiple take lanes on the same track, then mix them on the same track.

that's what makes reaper cool. doing it however is sort of a pain if you don't already know how.

other hosts i'm sure will be different.

however you have to do it, the next step:

play a polyphonic part as usual using your controller. make sure you record this using a soft-synth to test how it sounds. for example, load xhip and set it to the same number of voices on a similar sound and play.

once you have the polyphonic midi part, it depends on your host.

if you used the method as in reaper, the midi part should already have it's notes split into the right channels.

if you used cubase and it doesn't do this (i'm not sure if it does?) you'll have to re-process the midi part into a new midi part to split the channels. so you'd create a new midi channel, put the plugin in the instruments rack/whatever and select "notelogic" as the midi input. then put that new track on record and play the one you recorded before. that'll process it.

i think you can "process part" some how too but i don't have a copy of cubase so i can't test it for you. if so that would be the best because you could just click that and boom, the channels would be instantly split.

next step you have to mute or delete your old part/instrument and create one audio track per voice. (or in reaper, just use one and do multiple takes and mix them if you know how.)

put your synth on channel 1, or filter the midi part in some way to output only channel 1.

you may have to do this yourself. for example in most hosts there is a "split part into channels" feature which will split the part into 4 different "sub" parts.

mute the other channels and record the first 'voice'.

next, mute that and unmute channel 2, record the second 'voice'.

continue for 3 and 4, or as many as you want.

now you have a set of takes, one for each 'voice'. mix them and you have a polyphonic synth part.

the description took me twice as long as actually recording and mixing a 8-voice string sound 2-minutes long.

hopefully the second time you do it, it'll only take: voices X amount of time

God. That sounds super complicated. Ive got no idea what you are talking about. I guess I'd have to download the trial to get some idea.

Although I honestly have no idea what that post was actually referring to, I kinda get the rough impression that you are talking about a monophonic digital synth. Are you talking about creating multiple instances of a given soft synth and playing a voice per instance? If thats the case, you've completely missed the point. The original post is about an analogue synth.

He's trying to say:

1. record MIDI (with a prototype soft-synth sound) as if you had a polysynth

2. let the tool do magic to get a bunch of MIDI tracks (apparently on different channels; hence the MIDI-filter stuff) for mono synth

3. bounce them to audio one at a time (using your analog monosynth)

4. mix together

...

So this saves you the trouble of having to split chords to voices yourself.

Personally though, I'm not sure why manual voicing would be so complicated; I often do it with soft-synths too (eg duplicate a patch and set each one to mono-mode; useful for glides and some envelope tricks).

manual voicing is super complex if you're not just playing triads.

for example if you play something like superstition, how on earth are you going to manually voice that?

the voice allocator is the voice-allocation routine from xhip, a polyphonic synthesizer. it uses voice age and other conditions to find a suitable voice to use in the case that there are no free voices. one specific property is that it will always assign the same voice for the same note once a note has been assigned, unless that voice is unavailable.

so playing "d,d,d,d,d" will keep re-triggering the same voice. a very desirable property for certain types of sounds.

(it also allows you to prevent from stealing voices when it is just for a redundant note, highly desirable with few voices such as 2,3,4,6,etc)

because voices are allocated by age and counted "round-robin", if you set it to use 3 voices and play a chord with three notes using a chord plugin, the same three voices will always be allocated in the order the notes are sent. so if your chord plugin sends C,G,A,C and you have four voices set, the same voices will get used.

that means you can use a long fingered-glide on your analog monosynth and the whole chords will glide together if the chord plugin sends the new notes before stopping the old ones. normal glide (not fingered) will glide the set of notes in the chord smoothly together all the time.

it's of course optional but i only added the two options in the gui, was a really quick simple thing. i could do up a better gui with more options if anybody bothered to use it, but the whole concept seems to just go right over everyone's head.

seems nobody likes the complexity that is inherent in this kind of thing. you do get to learn it over time, same as playing a piano or guitar or anything else. you just have to get over that whole fret of fretting issue.

Took me a minute, but I got it. You're talking hardware synths. I sat here scratching my head, wondering why you didn't just switch your soft-synth to poly mode, or instantiate more plugins. The "Slim Phatty" didn't register.... need more coffee!!

Good tip, though. Ableton Live also has some pretty good pitch shifting devices. If you don't care too much about sound quality, you could also bounce down your synth's waveform and load it into a sampler.

Actually, having designed a reasonably intelligent voice allocator (eg stuff like prefer stealing sustain-pedaled voices over those actually still keyed) for my own synth projects, I do appreciate the complexity of writing one.

If I actually used polyphonic tracks more (and more specifically wanted to do so using analog monosynths) I'd probably appreciate something like your tool as well. On the other hand, I find that manual allocation isn't very hard (though arguably it's tedious) and can not only avoid all the corner cases of automatic allocation methods, but also enable expression that it simply impossible to get with automatic methods.

As for the chord-glide example: no, that most certainly is NOT what I would ever want to do.

it's just the fact that you can, really.

there are major differences between voicing done by a human and voicing done based upon a strict rule-set like applied by a computer program.

obviously if you're writing for multiple voices already, that's a completely different ball-park. for example the way you write for an orchestra is quite different from the way most people would write for a polyphonic keyboard part.

if you want to do polyphonic keys, the strict logic is really the only way to go once you get into any sort of situation that imposes any sort of questions. why bother asking or answering those questions when you can have all that done automatically and exactly to specification?

i prefer to use synthesizers in a multi-part way more than in a polyphonic way. they are two different ways of doing things though and one doesn't replace the other. you sure can manually allocate voices played polyphonic. it's like taking an improvisation played on piano and converting that to a jazz quartet though. sure it's possible, a heck of a lot of effort and pretty much no chance the quartet would sound like the piano.

http://soundcloud.com/aciddose/three-part-human
http://soundcloud.com/aciddose/three-part-notelogic

you can hear in this example of a human arranged three-voice composition. the processing applied by the voice allocator is actually perfect. completely perfect. it's well balanced and things are allocated in order.

only it doesn't sound nearly as good because in the human voicing parts are not split based upon strict ordering rules but rather they're split into three independent voices forming a harmony.

cg voice allocations obviously don't work well in that situation.

haha the buzzing + noise is actually a fan. that's what you'll get if you don't mute your mic track on the console

This doesn't seem like what aciddose is talking about. He's drifted off on a lot of other topics that I don't understand.

Is this what you are saying aciddose?

If so, again, why wouldn't you just split the notes up manually and record them separately? That's what I already do. Again, how does this xhip tool make this easier?

I would really like to know what aciddose is talking about.