• Register
  • Help
Page 1 of 3 123 LastLast
Results 1 to 10 of 23

Topic: sample modelling - harmonic alignement - how?

Share/Bookmark
  1. #1

    sample modelling - harmonic alignement - how?

    Hi Guys,

    I've listened with great interest to the Sample Modelling Trumpet demos, and I'm very impressed; especially how they've seemingly overcome the crossfade issues one normally gets on solo instruments when fading between a pianissimo and fortissimo sample.
    From the blurb on the site I can gleam this comes from harmonically aligning the samples..but the question is, how do you do this, and what software is used for it?

    Though I'm sure the guys from SM don't want to give any secrets away, I'd love to know where to start and try this on some of my library samples to get better crossfades on solo instruments

    Anyone got any tips?

    Cheers,

    Joe
    Joris de Man - Composer | Sound Designer - www.jorisdeman.com | Killzone 2 soundtrack now available on iTunes link

  2. #2

    Re: sample modelling - harmonic alignement - how?

    Joe,

    Thanks for your interest.

    Harmonic aligment is a tecnique I developed several years ago, which is now patented. Though theoretically simple enough, practical applications became rather complex, due to a series of algorithms and tricks, which were essential to get satisfactory results. Simply stated, harmonic alignment, as a mathematician could think of reproducing it, would not work

    I'd like to share with you, and other interested people, the principles behind this approach. However, I'm afraid that, at least as far as these further developments and tricks are concerned, this is going to remain a small secret. HA is only the start point of the samplemodeling approach. Much more has been developed and added to the original harmonic alignment technique. This is why I believe The Trumpet will be very difficult to overcome.

    I'm eagerly waiting to be disproved, since this will add fuel to the fire or "creationism".

    Friendly yours,

    Giorgio

  3. #3

    Re: sample modelling - harmonic alignement - how?

    Quote Originally Posted by Joris de Man View Post
    Hi Guys,

    I've listened with great interest to the Sample Modelling Trumpet demos, and I'm very impressed; especially how they've seemingly overcome the crossfade issues one normally gets on solo instruments when fading between a pianissimo and fortissimo sample.
    From the blurb on the site I can gleam this comes from harmonically aligning the samples..but the question is, how do you do this, and what software is used for it?

    Though I'm sure the guys from SM don't want to give any secrets away, I'd love to know where to start and try this on some of my library samples to get better crossfades on solo instruments

    Anyone got any tips?

    Cheers,

    Joe
    How about not crossfading at all? How would it work if you just switch the sample to the next velocity layer at a certain position? I'd suggest something like this, with common sense:

    Let's say you want to build a flute with constant vibrato, to keep it simple.

    -Record samples at x amount of dynamics.
    -Go through every dynamic level and find a spot where the vibrato is at 0-level of the "sine-cycle" and mark this spot
    -Create x dynamic layers and program it so that when you start using the modwheel, the volume just builds up and at certain point (when you're getting near to the next dynamic level), the sampler waits until the vibrato is at 0-level of the cycle and starts playing the second sample from the corresponding marker position as soon as possible.

    The advantage of this would be that you don't get any crossfade-buildup. Also, you could mark the other parts of the waveform too, so that the sampler doesn't have to stick with the current sample until the 0-level is reached, but it could also switch sample at other places of the sine-cycle.

    Maybe this is how people do already, I've never programmed any sampled instruments... but this would work particularly well with string ensembles, where you really want to hear the instruments clearly.
    Film Composer - www.juhanalehtiniemi.com
    My latest project: Our first indie pop -single, produced in Ableton Live, available also on iTunes and Amazon
    Mac Pro 6-core 3.33GHz

  4. #4

    Re: sample modelling - harmonic alignement - how?

    You would get a smooth volume curve but also timbral jumps.
    Samplemodelling is supersmooth in both aspects.
    rothermusic.de

  5. #5

    Re: sample modelling - harmonic alignement - how?

    Of course, I understand sample modelling way more advanced...
    Film Composer - www.juhanalehtiniemi.com
    My latest project: Our first indie pop -single, produced in Ableton Live, available also on iTunes and Amazon
    Mac Pro 6-core 3.33GHz

  6. #6

    Re: sample modelling - harmonic alignement - how?

    Giorgio,
    I wonder if it is possible with your advanced technology to do ensembles as well. Isn't sample modelling with the pure sound of a trumpet far easier to do than, say, a string section?
    rothermusic.de

  7. #7
    Senior Member Nigel W's Avatar
    Join Date
    Jan 2003
    Location
    Cologne, Germany
    Posts
    676

    Re: sample modelling - harmonic alignement - how?

    I would love to have a modelled clarinet.

    Is it any harder to model woodwind than brass?

    Nigel

  8. #8

    Re: sample modelling - harmonic alignement - how?

    Quote Originally Posted by Giorgio Tommasini View Post
    Joe,

    Thanks for your interest.

    Harmonic aligment is a tecnique I developed several years ago, which is now patented. Though theoretically simple enough, practical applications became rather complex, due to a series of algorithms and tricks, which were essential to get satisfactory results. Simply stated, harmonic alignment, as a mathematician could think of reproducing it, would not work

    I'd like to share with you, and other interested people, the principles behind this approach. However, I'm afraid that, at least as far as these further developments and tricks are concerned, this is going to remain a small secret. HA is only the start point of the samplemodeling approach. Much more has been developed and added to the original harmonic alignment technique. This is why I believe The Trumpet will be very difficult to overcome.

    I'm eagerly waiting to be disproved, since this will add fuel to the fire or "creationism".

    Friendly yours,

    Giorgio
    Hi Giorgo,

    Thanks for your response. I figured as much, and appreciate that you can't give all your tricks away.
    The most realistic 'other' trumpet I've heard is from wallander. It sounds pretty good, esp. as an ensemble, but doesn't quite have the same bite and attack as what I've heard from The Trumpet.

    Can you give any hints of where you would start with this kind of stuff, or is that still too close for comfort, so to speak?

    Best,

    Joe
    Joris de Man - Composer | Sound Designer - www.jorisdeman.com | Killzone 2 soundtrack now available on iTunes link

  9. #9

    Re: sample modelling - harmonic alignement - how?

    Joe,

    Can you give any hints of where you would start with this kind of stuff, or is that still too close for comfort, so to speak?
    One might think of Harmonic Alignment as of a kind of pitch matching between two or more sounds, but applied to each harmonic rather than just to the fundamental. This corresponds to a quite complex time warping.

    Crossfading among harmonic-aligned samples basically does the job of several concurrent time-varying convolutions, each loaded with the appropriate time-varying FFT coefficients for that particular note. Something not conceivable with the current technology.

    Harmonic Alignment carries out this task in a very efficient fashion.



    Justus,

    I wonder if it is possible with your advanced technology to do ensembles as well. ?
    Our technology was developed to allow continuous transitions across the dynamics (from ppp to fff), without the phasing issues which would otherwise occur with solo instruments. Crossfading ensemble samples is a standard technique, applied to several libraries, which is felt not to yield phasing problems. This might be an oversimplification though, since at the very least, the apparent number of the ensemble instruments actually doubles upon crossfading.


    Nigel,

    Is it any harder to model woodwind than brass? ?
    Not really. Just somewhat different. The real problem is that, creating a modelled instrument is a very time-consuming task.



    Thanks for your interest,

    Giorgio

  10. #10

    Re: sample modelling - harmonic alignement - how?

    Quote Originally Posted by Giorgio Tommasini View Post
    The real problem is that, creating a modelled instrument is a very time-consuming task.
    Which begs the question... so, when are those SM trombone, soprano sax, and clarinet libs going to be released? Oh, and the strings libraries, too?


Go Back to forum

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •