On Friday 18 March 2005 03:56 am, Emiliano Grilli wrote: > gioved?, 17 marzo 2005 alle 22:19:45, Juhana Sadeharju ha scritto: > > >> > At http://freepats.opensrc.org there is a mellotron sample in the > > >> > flac format. > > >> > > >> Here is a link to a draft SoundFont. No loop points were set, > > >> but looping is engaged. Approximately 12.3Mb. > > >> > > >> http://65.125.227.61/mellotronia.SF2 > > This one is made by John Check and contains all three sounds. > > > >As for looping, in the meanwhile a friendly guy from the italian > > >newsgroup it.comp.musica has done a looped version of the reed: > > > > > >www.viandanze.com/materiale/mello.sf2 (~3.7 Mb) > > > > Is this last sf2 based on the flac version? > > This one is the reed sound, cutted and looped by Paolo Ingraito. > > > How complete the original flac version is? > > There are the 3 mellotron sounds, sampled by a friend of mine who has a > music store and has sampled it before selling it. > > > Have the Mellotron tapes recorded from start to end? > > Unfortunately not, I think. And now the "real thing" isn't available > anymore. The samples are too short, and aren't complete (I don't know > how long were the original mellotron samples) Well.. the length is work-around-able but the key presses are inconsistent. There is a sort of aftertouch/velocity type effect from the mechanics of the instrument that he didn't account for which causes some modulation in the overtones. The effect can be duplicated with filters and envelopes. Since we'll never be able to duplicate the synchronization of the tones modulation (polyphonically) with straight playback, we will never be able to make a perfect "analog" (no pun intended) out of full tape loop samples anyway. I was considering taking short periods and then building on that with the SoundFont synth architecture. There is definitely a lot to work with in the samples. It's not beats, one can do more to make it nice than simple loops. Aaanyway... if you zoom in far enough there are repeating areas of flat amplitude several samples long that might be indicative of splices in the tapes or the masters from which they were cut. We won't be able to make a perfect rendition of that particular mellotron, but a good mellotron soundfont can be made out of what's there. > > > Have the end been faded-out after digitization? Is the end of audio > > the original end on the tapes or has somebody clipped the digitization? > > I don't think the notes were sampled for more than some second. > > > I'm only curious. Main thing is that the digitization has the start and > > as much as possible the sustained sound. > > > > I would suggest to use the original non-looped version for most proper > > sound. Not only because it is what Mellotron generated but also because A mellotron has actual physical loops with splices done by ear. Sweet irony. > > the non-looped sounds are better than the looped sounds. That part is a nonsensical statement without context. Here's what I did the other day when I took another pass at it (Don't wait for me to finish it): Decoded the original flac. Ditch the first 4-5 tones, they're duplicates. Cut the remainder into 3 sections, one for each timbre. Break the tones out to separate tracks/files Name 'em Normalize 'em (there's ~6db difference. Trim 'em Fade 'em if necessary If one looks at a spectragram it's pretty obvious that the whole waveform is being played (samples are long enough). The notes all start at varying places along the loops. There are issues with overtones and volume consistency over the course of the tone due to key pressure. The machine also needed lubrication or tape guide maintenance because there's at least one squeak. > > I agree, but that flac is all we have, and we have no more access to the > mellotron. But Paolo's work on looping is good (IMHO) for a 4Mb > soundfont, especially the flute > ( http://www.viandanze.com/materiale/melloflute.sf2 ) > > Best Regards,
