PipeChat Digest #2909 - Saturday, June 15, 2002
 
RE: Sampling versus real-time - another approach?
  by <cmys13085@blueyonder.co.uk>
 

(back) Subject: RE: Sampling versus real-time - another approach? From: <cmys13085@blueyonder.co.uk> Date: Sat, 15 Jun 2002 07:47:53 +0100   Hello,   Does anyone know what they are talking about?   Perhaps I am just stupid   Regards,   Colin Mitchell UK       -----Original Message----- From: "pipechat@pipechat.org" <pipechat@pipechat.org> on behalf of = "Adrian Nash" <adrian.nash@tesco.net> Sent: 14 June 2002 22:54 To: "PipeChat" <pipechat@pipechat.org> Subject: Re: Sampling versus real-time - another approach?   Hi Steve,   Thanks for your comments.   I see what you mean about modelling the synthesis to match the sample. = You=20 can do that with the PDA method, but in this case I would just do the=20 normal wavetable sampling if all I wanted was a reproduction of the=20 original sample - this is perfectly possible; the scheme supports either =   approach. The key point is that by first breaking down a sample into=20 partials it can be re-voiced to suit the voicer's requirements. The=20 original analysis of a sample isn't strictly necessary - it is a good=20 starting point - a template if you like.   I really would like to hear one of these new Yamaha synthesisers, it = sounds=20 very impressive. The "1 foot Sax" is a good example of how PDA would be =   used to make it sound better. If you took even the top note of a = Soprano=20 Sax then tried to shift it up to 1' pitch it would sound horrid and may=20 even "alias" due to high order harmonics going out of the frequency = range=20 of the synthesiser. However, with PDA, I would simply strip out these=20 troublesome harmonics or at least tone them down, this would have the=20 effect of rounding the sound and making it much more like what would be=20 expected. The problem with wavetable synthesis is that the harmonic=20 construction (and hence timbre) applies only at the actual note=20 sampled. The moment it is shifted in pitch, that harmonic construction = is=20 no longer valid. This is what makes samples sound "squawky" and why = more=20 keygroups are required. What I do is interpolate between two=20 keygroups. By phase locking two samples together and cross-fading = across=20 the keyboard, you get a gradual shift of timbre between the samples = rather=20 than an abrupt change.   Reverb is not really my field. I think that in the end, there is a = limit=20 to what can be achieved with electronic reverb, it is still probably = best=20 to provide a digital organ with a large number of speakers and let the=20 natural acoustics do the "processing". However, in a home situation for =   practice purposes, this is not usually practical. I have found that=20 exploiting the 3D hardware mixing capabilities of the new breed of sound =   cards does go some way to improving the "mix" of pipe choruses. The=20 problem is that a "flat" mix that does not include effects such as=20 time-delay and phase changes not to mention the reverb aspects can = suffer=20 from the A + B equals C phenomenon ie, two distinct waveforms when mixed =   together are not heard as two separate waveforms anymore but a sort of=20 hybrid that sounds like neither.   Velocity response? Yes, you can change the timbre with repect to = velocity=20 so you can vary the amount of chiff on an un-nicked pipe for example in=20 real-time. The problem is, you need a velocity-sensitive keyboard which =   most organ consols are not. The trick is to layer two phase-locked=20 waveforms together with different velocity-attack characteristics. As = the=20 velocity is increased, one waveform will start to dominate the other, = for=20 example bringing out more 3rd harmonic and hence chiff.   Sorry for *another* long reply! Best Regards Adrian     At 15:06 14/06/02 -0500, you wrote: > > Adrian Nash wrote a rather lengthy response to Cheryl at Copeman = Hart: > > > > It is astounding just how powerful a 2GHz PC is these days. The = main > > limitation seems to be the operating system rather than sheer = processing > > power in these real-time applications. Even so, Windows doesn't do = too bad > > a job considering I can download my email or my wife writes a letter = in > > "Word" whilst I am playing my heart out with all the stops drawn - = through > > headphones you understand, my wife is tolerant but not *that* = tolerant! > > > > This is a "home grown" project borne out of an interest in this kind = of > > thing and a lack of budget for a really good digital organ. I've = found > > like many subscribers, the cheaper end of the digital organ market = just > > doesn't cut the mustard but yet, who has upwards of $10,000 to buy a = really > > good one? Not me! I would be very interested in others comments and > > suggestions. > > > > > >Adrian I'm curious about a number of things. From what you describe you =   >model the synthesis to match what you sample. I don't see how that is=20 >significantly different from sampling, you just use a different route = to=20 >get the same result. > >I have a Yamaha synthesizer that does virtual modelling. It has models = of=20 >Saxophone or trumpet and then calculates what it would sound like with=20 >variables for breath pressure and lip pressure. It's fairly realistic = but=20 >the resulting sound is very raw, imagine a Sax at 1 foot. It can be=20 >softened by mixing in a sample and/or adding ambience. I'm a bit of a=20 >perfectionist on reverbs because my end result is a recording not just=20 >personal enjoyment. Really good reverbs are very processor intensive. > >I also wondered if your synthesis included velocity response and = attempted=20 >to model tracker action? That would seem pretty cool, but I'd have no = idea=20 >of how to do it. > >Thanks for your post it was very interesting. > >Steve Chandler > > >"Pipe Up and Be Heard!" >PipeChat: A discussion List for pipe/digital organs & related topics >HOMEPAGE : http://www.pipechat.org >List: mailto:pipechat@pipechat.org >Administration: mailto:admin@pipechat.org >Subscribe/Unsubscribe: mailto:requests@pipechat.org     "Pipe Up and Be Heard!" PipeChat: A discussion List for pipe/digital organs & related topics HOMEPAGE : http://www.pipechat.org List: mailto:pipechat@pipechat.org Administration: mailto:admin@pipechat.org Subscribe/Unsubscribe: mailto:requests@pipechat.org