Posted
Another 8-bars sample.
The marimba part is a "distillate" of different overlapping very short figures recorded from my midi keyb, then reduced to a texture as simple as possible using Synfire's tools ("drop repetitions", to say one) in multiple passages.
To shape the melodic profile I simply moved up/down the figure blocks with the arrow keys.
My goal is to create clarity in complexity over much more interesting harmonies of these common chord turnarounds and, hopefully, write using orchestral libraries and letting Synfire find simple solutions to complex problems.
Otherwise the assisted composition software serves to no purposes :-)
marimba.mp3
Thu, 2019-03-21 - 13:10 Permalink
Thanks again for sharing your experiments. Here's what I did with your file:
- For polyphonic counterpoint-like figures, it' always best to group segments melodically. The vertical chords will "just happen" by accident. What a listener is following are the melodic lines (or what is perceived as such), so these are most important to not get torn apart (= grouped). I tried to do that w/o shifting any note symbols. Now if harmony is changed, the melodies will remain coherent.
- The extra full chords instrument didn't really fit the marimba, as nobody would be able to play the resulting polyphony on a real instrument, so I muted them.
- The playing ranges were all set to 'upper'. In this case adjusting the playing range of the instrument makes sense, so you can use 'middle' range for the figures.
marimba.mp3
Thu, 2019-03-21 - 17:30 Permalink
It was the first time I got from Synfire kinda "plot" of interesting parts using the embedded tools. The idea was to "dig" from the inside out a melodic form, but I immediately abandoned the project to try others.
Perhaps one would think of introducing in Synfire some post-production parameters as implemented, for example, in Riff Generation by InSession Audio? (not necessarily the same parameters, of course!)
Thu, 2019-03-21 - 18:47 Permalink
Yes, generative stuff is what we are currently working on. Since Synfire already knows everything around Harmony, generating Figures is relatively easy.
It will however not be some step sequencer like random stuff. It will be a structural thing that builds up from rules and that uses motifs and transformations.
Sat, 2019-03-30 - 18:35 Permalink
It will however not be some step sequencer like random stuff. It will be a structural thing that builds up from rules and that uses motifs and transformations.
In my opinion, what is missing in the panorama of assisted composition software is a tool that operates on structures, rather than on individual notes.
For example, it would be nice to be able to define:
- the length, or total time duration, of a musical phrase
- its melodic and/or rhythmic climax, and in which voice/instrument it should be found
- occurrences of a specific note, or of a rhythmic figure, or of a chord
In simple words, it would be nice to be able to "paint" the piece in general characters, at least in its general lines, using a "larger brush", rather than the small ones currently provided by the mainstream daws and generative software currently on the market.
Sun, 2019-03-31 - 14:36 Permalink
In simple words, it would be nice to be able to "paint" the piece in general characters
I see the point. The technology that could read a composer's mind however is not yet invented. For any genre and style, there are very different techniques and instrumentations to achieve a particular "general character".
The only viable way to have parameters associated with emotions, listener experience and other subjective concepts ("tension", "sadness", "climax", "danger") is to have style templates that meticulously follow the idioms of a particular, well-known style. Then, yes, these templates could have pre-manufactured variations controlled by parameters. It certainly would be a great selling point in the short term, but this type of instant clichés get old very quickly.
Logic X Drummer does this parameter-driven style variation very well. But a great drum track only gets you so far in making a great song.
Sun, 2019-03-31 - 17:00 Permalink
I see the point. The technology that could read a composer's mind however is not yet invented.
I've never really thought about it, but I suppose you can also parameterize the musical macrostructures regardless of style: after all, a climax is a climax. Whether it's a rhythmic, dynamic or melodic climax or all of them at once.
I also bought Pizzicato, a French software that at first glance looks more like an arranger than a composer assistant.
I'll start studying it after Rapid Composer and Orb.
Let me think about all these intriguing things, the last word is not yet said... :)
Mon, 2019-04-01 - 13:37 Permalink
It's definitely interesting for the new music-as-a-language generative features we are experimenting with currently.
Exactly, Andre. I don't expect, nor hope, that a software will replace me.
But I hope it will help me to explore new possibilities more quickly than a human being can in a lifetime :)
