Sample library and synthesis musings

[Consul]

Probably the most important contribution I see myself making to the new project is the creation of sample libraries and associated patches. Lately, I've been really inspired by the videos of the new Spectrasonics synth, Omnisphere:

http://www.spectrasonics.net/omnisphere ... plore.html

There are also videos on YouTube where an entire presentation at NAMM was uploaded:

http://www.kvraudio.com/forum/viewtopic ... 21#2917621

I like the idea of collecting oddball samples and using them to craft sounds. I'll probably try my hand at building some cool instruments to sample myself. I also have a friend with some nice old gear, like an Arp Solina, a nice big Hammond console organ (tonewheel), a Prophet 5, and a Rhodes, among other things, that he's willing to let me sample. I'd also thought about getting my hands on some old tube radios and seeing what sounds I can make happen. The sky's the limit!

For more down-to-earth stuff, at the school I go to, there is a nice grand piano and a harpsichord in a decent room that I might be able to get enough time on to sample. I also want to do a nice, big drum library, but I have a special plan for that. The idea here is to record hits for each drum bone-dry and with a nice, flat mic, then use impulse responses of mixers and rooms to create a cohesive sound (the Larry Seyer approach, basically). It still involves getting my hands on a drum kit, though. Maybe someone local will let me borrow a nice one for a bit.

I can probably also get a music student or two to help me out with some various other samples, like violin, woodwinds, etc. I have some special ideas for re-synthesizing those timbres that I'd like to try out, actually. This would involve creating single-cycle samples of each note in the range of these instruments, which are then split across the frequency spectrum into different single-cycle loops. The idea then is to control each component of the total sound separately with envelopes, based upon an analysis of various articulations of the real instruments. It's hard to explain in just a paragraph, so I'll go into more detail later if anyone wants me to.

So for me, if the new engine can allow me to do all of this, I would be happy.

[lowkey]

Probably the most important contribution I see myself making to the new project is the creation of sample libraries and associated patches. Lately, I've been really inspired by the videos of the new Spectrasonics synth, Omnisphere:

http://www.spectrasonics.net/omnisphere ... plore.html

There are also videos on YouTube where an entire presentation at NAMM was uploaded:

http://www.kvraudio.com/forum/viewtopic ... 21#2917621

I like the idea of collecting oddball samples and using them to craft sounds. I'll probably try my hand at building some cool instruments to sample myself. I also have a friend with some nice old gear, like an Arp Solina, a nice big Hammond console organ (tonewheel), a Prophet 5, and a Rhodes, among other things, that he's willing to let me sample. I'd also thought about getting my hands on some old tube radios and seeing what sounds I can make happen. The sky's the limit!

For more down-to-earth stuff, at the school I go to, there is a nice grand piano and a harpsichord in a decent room that I might be able to get enough time on to sample. I also want to do a nice, big drum library, but I have a special plan for that. The idea here is to record hits for each drum bone-dry and with a nice, flat mic, then use impulse responses of mixers and rooms to create a cohesive sound (the Larry Seyer approach, basically). It still involves getting my hands on a drum kit, though. Maybe someone local will let me borrow a nice one for a bit.

I can probably also get a music student or two to help me out with some various other samples, like violin, woodwinds, etc. I have some special ideas for re-synthesizing those timbres that I'd like to try out, actually. This would involve creating single-cycle samples of each note in the range of these instruments, which are then split across the frequency spectrum into different single-cycle loops. The idea then is to control each component of the total sound separately with envelopes, based upon an analysis of various articulations of the real instruments. It's hard to explain in just a paragraph, so I'll go into more detail later if anyone wants me to.

So for me, if the new engine can allow me to do all of this, I would be happy.

+1

[Consul]

You know, I keep forgetting I have an Optigan downstairs in the basement. It was in a quasi-working state last I checked, as in, I could get it working long enough to sample it. Yet another source of odd sounds...

[dahnielson]

You know, I keep forgetting I have an Optigan downstairs in the basement. It was in a quasi-working state last I checked, as in, I could get it working long enough to sample it. Yet another source of odd sounds...

Wow. That's an "instrument" whose existence I was totally unaware of despite the fact I engrossed myself in synthesis and sampling history last year. Sure, it may it may just have had the same influence on development of contemporary music as the Fisher-Price PXL-2000 had on the advancement in cinema. But the Wikipedia article about it was worth reading.

Apparently there is an official samplings CD selling for $199 (in case you have everything or decide not to go for the Kirk Hunter's Emerald Orchestra for $325, Sampletekk's Black Grand for $139 or some of the Westgate Modulars for even less money... ).

[Consul]

I'm thinking about how to go about sampling that grand piano, and I'm not confident I can hit keys evenly enough to get 32 layers so consistent across the entire 88-key range without building a robot (of course, that would be cool in and of itself). That means I'd likely have to find an experienced pianist and get him or her to join me for a very boring all-weekend session. Then I'd want to do some oddball things, like sample plucked strings.

I have an old upright piano here at home that I suppose I can get some practice sampling. It actually has quite a nice sound, though it does need some mechanical help (two of the keys aren't functioning right) and a tuning. Who knows? Maybe "Old Upright Piano" might become a popular library.

[dahnielson]

I'm thinking about how to go about sampling that grand piano, and I'm not confident I can hit keys evenly enough to get 32 layers so consistent across the entire 88-key range without building a robot (of course, that would be cool in and of itself). That means I'd likely have to find an experienced pianist and get him or her to join me for a very boring all-weekend session. Then I'd want to do some oddball things, like sample plucked strings.

I was discussing the piano velocity layer question in this thread (among other things):

http://www.northernsounds.com/forum/sho ... hp?t=58730

[cuse]

The high quality piano libraries are usually not made by hitting the keys of the piano manually. Instead they're using machines to do that, which allow to set the force, thus the velocity with wich robot is going to hit the keys, very precisely. This is the fastest and probably best way regarding quality to make a good piano library. But you know..... you need somebody to make / rent you a little hydraulic system to do that. I think there's not really an alternative to an hydraulic solution when choosing to go the robot way. Because all other ones (e.g. pneumatic or electrical servos) would create noise which definitely should not accompany your samples.

Another, more cheaper way is to to do trigger the keys manually, trying to do that in all kinds of velocities hundreds and thousands of times for each key, and then select the appropriate samples afterwards. The idea hereby is that the more time you hit the same key, the higher the chances that you end up with having your samples for e.g. all 128 velocities (or how much splits you want to achive). There is already software which automatically analyzes and picks the right samples for you out of the set of samples you give to the app. So it compares the energy of all samples you provide and then decides which would fit to the respective zones you want to have. But ... it is really like it sounds: a looooooooong process. Because each time you hit a key, you have to wait and sample it until it fully deceased.

But maybe you know some engineering student which can help you out with a hydraulic solution. Because it's really not that hard to build such a little robot "finger".

[dahnielson]

Per "Worra" Larsson (of Sampletekk whose Yamaha C7 have 93 velocity layers) said that he once tried a mechanical system but it made too much noise for the samples to be useful, instead, he use custom software that let him sample pianos in 2 dB steps.

[Consul]

Well, given that I *am* a mechanical engineering student, I might be able to think of something. NI's Akoustik mentions the use of a mechanical aid to play the notes, and given that their library is only white keys, I'm willing to bet their robot couldn't play the black ones.

The harpsichord should be easier. Of course, this is all contingent on my being allowed to touch these instruments by the music department.

[dahnielson]

Here's some inspiration for you then:

http://www.qrsmusic.com/pianomation_why.asp

(Solenoids are fun )