Advanced Max: FFTs, Part 4


    Now that we've got the basics for understanding FFT processing in Max, let's pull it all together and make something wonderful. This week's 42-minute video pulls pieces together from previous installments of this series to create a Max for Live device that does independent processing of each band of audio to produce spectacular results!
    All the tutorials in this series:

    Advanced Max: FFTs, Part 4


    • Apr 11 2017 | 12:31 am
      Thanks once again for these and do please, if possible, keep this sort of content coming!
      Peace.
    • Apr 14 2017 | 1:42 pm
      thanks tim as ever. the 'advanced' tutorial series in general is fantastic.
      however, i am slightly perplexed by this #4 tut, and have to be a whinger here...
      1) unless something wonderful has been developed since the last time i checked a few years ago, @parallel for poly~ does not work in M4L - nor do most of poly~s useful features, such as dynamic loading etc. poly~ is half crippled in the Live environment 'cos of each track being on a thread etc.
      2) i really do comprehend pedagogy and ordering tutorials to introduce concepts and learn better through doing, but is it really actually useful to make a poly~based multiband delay that takes up 30% CPU in live (20% in max 'cos @parallel works there)? especially when there are very well known and far better and just as easy to comprehend techniques already out there, such as the olivier pasquet / manual poletti "Max Spectral Delay" that has always come with M4L, is far higher quality and uses only 3 or 4 % cpu ? :-)
    • Apr 15 2017 | 10:18 am
      Yeah, 30-40% cpu is a killer, though I like the vertical GUI—it reminds me of FreqTweak! All we need is an LFO now.
    • Apr 15 2017 | 10:27 am
      @stkr
      good points and observations; may I make one? Sometimes reinventing the wheel is useful as a didactic exercise/approach, so that others can build other wheel-types and non-wheels. Although, as I began watching this, my immediate thoughts centred on CPU demands and M4L restrictions.
      Great series, with much to learn. Kinda got lost around the framecount quantization, but that's my failing.
    • Apr 17 2017 | 3:41 pm
      All fair points, and I appreciate the feedback!
      It can certainly be perplexing trying to balance legit concerns of advanced users (e.g. cpu load and resource management) and clarity for folks aspiring to become advanced users. I certainly don't always get it right.
      For what I was trying to accomplish, Manuel's awesome spectral delay device was a bit too magical. Because of the way the delays are managed with the shared tapin~/tapout~ objects it is almost impossible to explain without drawing lots of elaborate illustrations on a white board.
      I was also hoping to provide a framework where other types of spectral processing could be explored, not just a delay. The suggestion here for adding band-specific LFOs is a great example.
      I'm embarrassed to say that I forgot that MFL doesn't support poly @parallel 1. I developed the patcher initially in Max and that feature makes a huge difference in Max. Hopefully it is still useful for folks to know about the feature.
    • May 08 2017 | 8:48 am
      Thanks bunches for this series! The spectral delay eats 4% CPU on my laptop, btw. I hope you can add more to this series, like gen~ stuff. It is much appreciated!
    • May 13 2017 | 9:22 am
      Great Tutorial! Definitely that @ parallel 1 was a good insight. That cmnd-j shortcut slipped my mind, thanks again.
      Wouldn't it be great to have another way to initialize a new saved poly~ instead of always change the name, or delete and undo?
      Looking forward to the next one, very good level!