Not signed in (Sign In)

Welcome, Guest

Want to take part? Sign in or apply for an account below

Vanilla 1.1.4 is a product of Lussumo. More Information: Documentation, Community Support.

    • CommentAuthorrob
    • CommentTimeJun 7th 2008 edited
     
    This is my first HighC work. It borrows its organizing principle from Xenakis' 1954 piece "Metastasis."

    I think this would benefit from some post processing.
    • CommentAuthorthbb
    • CommentTimeJun 12th 2008
     
    Thanks rob for this contribution. I can appreciate it's been a bit of work to do that in HighC. I'm proud there are people doing this with HighC...

    There is a limited amount of post processing that can be done in HighC: adding echo or reverb by selecting a group of sounds, duplicating it and lowering its level (and repeating the process). The attached file shows you a sampler of what can be done in this respect.
    • CommentAuthorrob
    • CommentTimeJun 13th 2008
     
    You can see a spectral analysis at this URL:
    http://stashbox.org/137954/study-after-xenakis-spectrum.PNG

    I'm very impressed. You can clearly see the harmonics making up the triangle wave, and you see quite clearly the linear frequency trajectory in time as described in the bug thread.
    • CommentAuthoradmin
    • CommentTimeJun 13th 2008 edited
     
    Indeed, you even see that in the high range (fourth section), the triangular waves generate heavy aliasing (blueish area in the 0-1kHz range)! I hope to be able to filter the "waveshape" waves one day so that aliasing is less pronounced. Still, seems like your sample file is still in 22kHz. Using "High Quality" rendering mode should reduce this artifact (at the expense of speed)...

    If you want to get rid of aliasing (some don't), I'd suggest using the clarinet "additive" waveform, or some other similar waveform.

    Thanks for the image. For me, it also shows that if you wanted to come back and edit such a piece with a "bitmap-based" tool like Coagula and metasynth, it would be quite painful to do (preserving the harmonics and amplitudes...)...

    About your "etude after ligeti": I really like it. I think that with a properly spread "noise>bandwaveform" waveform, and importing the midi file in HighC, it should be quite easy to reproduce with good accuracy.

    Of course, stereo is still lacking in HighC.. that has to wait a bit, but I intend to provide full spatialization in the same time as stereo...

    Still about Ligeti, did you know this video? It's an inspiration for improving HighC... For instance, I could allow replacing the visual representation of a pattern by any kind of drawing, at the will of the composer...
    • CommentAuthorrob
    • CommentTimeJun 13th 2008
     
    Yes, that video gave me the impetus to pursue this avenue seriously. In that case it's sort of the inverse of HighC (ChgiH?) in that the score was created after the realization of the work in order to depict the work. I commented elsewhere about a set of complementary transforms from sound to image (like an FFT and IFFT).

    The depiction of reverb as a sort of background cloud in that score really appeals to me.

    Returning for a minute to the idea of multiple views like a set of architectural drawings depicting various aspects of a work--if you could toggle to a spatialization view that depicts the same sound objects in a stereo (or surround) field, and then move the sound objects in the field, that sort of UI could fit the overall HighC paradigm pretty well. The trick is knowing which object you're moving in the new view I suppose. Timeline remaining consistent should help.

    In several resources on mixing, the idea of a 3-dimensional soundstage is used, where the left-right position is governed by pan control, the front-back position by wet/dry mix on the reverb (with dry corresponding to up front), and top to bottom corresponding to register (it's suggested that psychoacoustically we localize lower register sounds down, but I've never really read any original research to support that).

    So I don't know exactly what you intend as full spatialization, but these are some ideas.

    As for your comments above on aliasing--it may also be a side effect of the top frequency I selected for the FFT. I believe that I rendered in HQ, and I also believe (I'm away from the machine right now) the file was identified as 44.1kHz sample rate when I brought it into audacity. I was in a hurry though, so I'll verify those beliefs.

    The HQ render sounds vastly better to me. Dropping a bit of stereo reverb on it also makes a lot of difference. I'll post up a post-processed version later today.
    • CommentAuthorrob
    • CommentTimeJun 14th 2008
     
    This HQ render has stereo reverb added, plus a small amount of audio pixie dust.

    http://stashbox.org/138389/study-after-xenakis.mp3
    • CommentAuthorrob
    • CommentTimeJun 20th 2008
     
    I just listened to this at full volume for the first time. That last gesture just rips you apart when the glissandos cross right there near the peak of human hearing sensitivity.
    • CommentAuthoradmin
    • CommentTimeJun 22nd 2008
     
    I finally got to listen to it properly last night. It sure sounds like Xenakis' at its greatest.

    Shows also clearly the need to post process HighC's output: the reverb at the end of the second sound wave is really needed and adds compared to the silence of the raw upic file.
    I also realize something a few have told me about it before: aliasing (in the last gesture, around 1:20) actually doesn't hurt at all. In the contrary, it sort of "prepares" the hear for the upcoming crescendo.

    I have some ideas on how to create several effects graphically, but that will also have to wait.
    Surely, HighC will be available as a VST plugin before effects are implemented (it's probably more useful).

    I suppose you're OK for me to put it on the samples page?
    • CommentAuthoradmin
    • CommentTimeJun 22nd 2008
     
    As for spatialization, I will open a separate thread to start discussing it.
    • CommentAuthorrob
    • CommentTimeJun 22nd 2008
     
    Of course, yes.

    I'm working on expanding this sketch into a proper composition. I see now that I can use the tags to prepare for deriving two (or more) separate files to make a spatialized version from the single upic score file. Then I can render the channels separately and pan accordingly in my DAW.

    I'll also add that conversion to MP3 removes the spectrum above 16kHz. Not that it's an enormously significant part of the signal, but the quality of the 24-bit wav render is quite a bit better than the mp3 reflects.
    • CommentAuthorrob
    • CommentTimeJun 22nd 2008
     
    >HighC will be available as a VST plugin

    If that were the case, then the full range of audio processing becomes available to HighC users without you having to implement any of it. Seems like a worthy architecture goal.