VCSL

[bigcat1969]

Hey Sam. I've been fighting off something nasty that's been floating around here and sleeping a lot, but when I'm a bit better I'll take a closer look at the samples.

You warned me early on in the thread that names and other things might change and that's cool. Consistency in naming is useful and sometimes surprisingly tough to achieve.

Hopefully the sfz thing works and peastman becomes the king of sfz creation. If it becomes easy enough / worth doing there are plenty of Kontakt instruments that it would be nice to see an sfz option for.

[peastman]

I just sent a PR with the final set of SFZs, so all instruments are now done.  There's probably a lot of room for improvement though.  Any feedback on them would be great!  I already have some ideas on things I want to change.

Now that I've actually spent time working with every instrument, I want to say this is an amazing sample collection.  Sam, you're incredible for creating this.  I can hardly imagine the amount of time and effort that must have gone into it.

Quote:Paul has been doing a lot of heavy renaming of sounds

Peter, not Paul.

Quote:although one of the timpani I think still has no pitch names/tuning

Yeah, the timpani still needs work.  "Timpani 2" especially is complicated because it includes notes from a bunch of different timpani with overlapping ranges and no consistent tuning.  I need to go through them and select a subset of the samples I can arrange into a single coherent scale.  There's plenty of material to work with, but it will take some manual curation.

Quote: Haven't gotten it out yet, but mapped one of the Kiowa Pianos for Maize. Forgot how much fun mapping pianos was!

If you haven't already, check out the Steinway and Knight pianos.  They're my favorites, beautiful instruments with very clean, consistent samples.

[Samulis]

(05-05-2018, 06:35 PM)peastman Wrote: Peter, not Paul.

I knew it was one of the disciples, apologies. 

Just pushed the ocarina fixes and the new official logo for the project-
   

It's intended as a symbolic reference to the "analog-to-digital" nature of sampling.

Here's the full fancy logo-
   

[Samulis]

Peter, I just checked out the latest versions. Some really great stuff, especially with some good reverb added! The saxes and harmonicas are an absolute joy to play.

I found a few things-
- The Dan Tranh was not relabeled right or perhaps needs some serious tuning adjustments. The B's are all an octave too high. It also cuts off notes on release when it should sustain openly.
- Some of the strumstick notes are quite out of tune (e.g. Strumstick_Finger_Str2_Main_D3_vl3_rr1.wav). Not sure if you can fix this on your end; I thought I had the samples tuned.
- The pianos sound much more realistic with a release of 0.4 (I have gone ahead and made this change).
- Starting on E6 (C3=60) and up, there are no more releases on the piano and all keys should be given a long release of something like 10.0. This *can* be done by hand, but might be better to add to the script, since it is a feature of virtually all modern pianos.
- It appears Steinway B is actually C4=60. I don't know why or how (I swear I thought it was C3=60), but it is.
- It is HIGHLY suggested that Steinway B should not be tuned. In fact, if you want to tune a piano, you should apply exactly identical tuning across all samples of the same note name (e.g. all the E6 samples should be offset together by the same amount with tune=). The best place to decide this offset from would be the middle velocity layer, or in absence, the lowest velocity layer. As the piano is hit harder, the first few waves after the hit of the hammer may appear to be at a higher pitch. This fools digital tuners into thinking the louder notes are sharp, when after half a second they fall down to the pitch they *actually* have. The same issue occurs on all harps, guitars, and other plucked instruments (including the plucked psaltery) which have dynamics, so this may be part of that strumstick tuning issue if there's any tuning on that.
- There is an errant extra RR in the English Harpsichord on A#4 in the Lute patch. I'll remove this sample in the main branch. There are also a few notes which have odd resonances which I will need to check (e.g. F#3) in the same. I also don't think the tuning belongs on the English Harpsichord (or any other harpsichord), it sounds out of tune. All of the harpsichords should be perfectly pre-tuned. The number of instruments which aren't pre-tuned and need tuning scripted in should actually be quite low, since it's a standard part of the process for cutting most of them. The only generalized exception is lamellophones like the mbiras and kalimbas, which don't comply to western tunings.
- For clarity, it might be best to rename 'Harpsichord, Flemish - High' to 'Harpsichord, Flemish - 4' ' and "Low" to "8' ".
- It's may even have to be manually done, but the Plucked Idiophones will need to be tuned.
- The tuning in Bell Tree - Individual (D5 +33c) was incorrect, so I commented it out.
- Having a tight release on Claps made no sense, so I made it 2.0.
- The offsets in Gong2 seem incorrect and cause very strange stuttering behavior in ARIA.
- Note to self: too much background noise in Siren and Shaker, Small, fix weird noise in Tamb2 Shake rr3.
- Tamb1_Shake_rr1_Mid can be have even more cut off the front. I think I will instead just go and edit the sample itself.
- Tambourine 4 - Legacy's "tamb2_rr1" is louder than the other 3rr. It might be better if we rename/re-purpose it as an upper velocity layer.
- Tubular Bells 2 are an octave too high (i.e. they're C4=60).
- Tubular Glockenspiel is at least one octave too high.
- Definitely do not tune the mallets (xylo, marimba, glock, vibes, tubular bells, etc.) as these all have fixed pitch at time of creation and do not really change pitch over time unless horribly broken.
- Xylophone is one octave too high (i.e. C4=60). It's one of those older instruments...
- Changed release of Ocean Drum to 4.0. In reality, the darned things are almost impossible to stop, haha.
- Bass Drum 3 appears to be a little too loud, I managed to make it clip. Maybe knock 2 dB off on average.
- Legacy Snares all appear quiet. Suggest adding 6 dB.
- Timpani1 appears an octave too low, but it sounds fantastic.
- Attack of 0.1 on soprano recorder sounds unrealistic (the instrument is supposed to have that chiffy start), so I removed it.
- Technically the organs should all have no velocity-sensitive dynamics, instead they use typically some form of swell shutter.

The Hi-Hat should follow the GM map spec-
http://computermusicresource.com/GM.Perc...eyMap.html
Close=44
HitC=42
HitO=46

HitLoose & HitOC can probably be mapped to 43 and 45 then, at least for now, although they don't have an official home in the GM spec.

Similarly Suspended Cymbal 'hit' could possibly go on the note for Crash Cymbal and 'hit_stick' can go on the note for Ride Cymbal, with Bell on the Crash Cymbal 2 spot or whatever... or we can just leave it as is (although it would be better if we eventually moved the hits to the bottom of the range and put the effects higher up; that's a typical mapping configuration). Right now the percussion are sort of mapped wherever (I think they were mapped just alphabetically?), so that'll need to be fixed at some point so they have a more logical and coherent order. Likewise we'll need to come up with a plan for snares. Generally I try to put the most common/important notes on the white keys, but I'm open to ideas to try to make it more easy to map with the script.

Edit- just pushed my sfz changes to the sfz branch. Will see about sample edits tomorrow or the day after as time permits.

[peastman]

Thanks!  I'll get to work on those changes.

By the way, I found one other incorrect sample.  On the Yamaha piano, Upright1_Sus_C0_vl3_rr2.wav is actually D0.

Have you thought about how to make these files accessible to users?  Right now they're hidden where few people are ever likely to find them.  If you browse the repository, you won't see them.  If you download a zip of the repository, you won't get them at all.  They aren't even mentioned in the readme.

[Samulis]

(05-12-2018, 03:56 PM)peastman Wrote: Thanks!  I'll get to work on those changes.

By the way, I found one other incorrect sample.  On the Yamaha piano, Upright1_Sus_C0_vl3_rr2.wav is actually D0.

Have you thought about how to make these files accessible to users?  Right now they're hidden where few people are ever likely to find them.  If you browse the repository, you won't see them.  If you download a zip of the repository, you won't get them at all.  They aren't even mentioned in the readme.

Thanks, I'll check that piano sample out. Very weird that it's a D, I don't recall sampling any D's on that instrument! XD

Might be the samples got tuned wrong after cutting...

Well, currently they can be accessed easily by clicking 'branch' and selecting 'sfz'. I will probably put instructions in the readme for this as you suggest. At some point, I may put up a handy .zip folder, but we need to start versioning (#.#.#) our sfz releases then for that to make sense.

The ideal case atm is for people to actually sync to it with Github Desktop, that way when we change 4 files, they get changed and the user doesn't have to download a patch and apply it or, even worse, download the entire thing again. Unlike VSCO 2 CE, it looks like we're going to be updating this every week/other week or so for the foreseeable future, with possibly even more contributions from other folks too.

Edit- in this case, maybe a download video walkthrough makes sense!

[Paul Battersby]

(05-12-2018, 09:15 PM)Samulis Wrote: I will probably put instructions in the readme for this as you suggest.

Hey, Sam, in the instructions, it would also be helpful to tell people what Aerophones, Chordophones, Electrophones, Idiophones and Membranophones are so people know where to look for the various instruments. To me, these labels are an obfuscation (even if they are actually the most accurate words to use for someone familiar with them) and I have to dig into each sub directory to know where to look for anything. I suspect others might have the same problem.

[Samulis]

(05-12-2018, 09:45 PM)Paul Battersby Wrote:

(05-12-2018, 09:15 PM)Samulis Wrote: I will probably put instructions in the readme for this as you suggest.

Hey, Sam, in the instructions, it would also be helpful to tell people what Aerophones, Chordophones, Electrophones, Idiophones and Membranophones are so people know where to look for the various instruments. To me, these labels are an obfuscation (even if they are actually the most accurate words to use for someone familiar with them) and I have to dig into each sub directory to know where to look for anything. I suspect others might have the same problem.

Will do! I just used these because we're going to be putting some really difficult to classify instruments in there as time goes on- stuff that isn't even related to an orchestra, where traditional WW/Br/Str/Perc lines get crossed and twisted.

I might try to do a little chart with pictures, I think that would be even easier.

[peastman]

Quote:At some point, I may put up a handy .zip folder, but we need to start versioning (#.#.#) our sfz releases then for that to make sense.

An easy way of doing that is to create a "release" for it.  You specify the branch/revision, give it a version number, and then github provides a direct link to download a zip of that revision.

Quote:it would also be helpful to tell people what Aerophones, Chordophones, Electrophones, Idiophones and Membranophones are

Yeah, I had to look those up!

---

Quote:Starting on E6 (C3=60) and up, there are no more releases on the piano and all keys should be given a long release of something like 10.0.

You don't really mean 10 seconds do you?  I just tried this, but it sounds like the key is never released.  The damper should silence the string in a fraction of a second.

Quote:Attack of 0.1 on soprano recorder sounds unrealistic (the instrument is supposed to have that chiffy start), so I removed it.

The reason I put that on is that a few notes have pronounced squeaks at the start that make them stand out from the rest of the instrument.  But I can move it to just those notes instead of the whole instrument.

[Samulis]

(05-12-2018, 10:24 PM)peastman Wrote:

Quote:At some point, I may put up a handy .zip folder, but we need to start versioning (#.#.#) our sfz releases then for that to make sense.

An easy way of doing that is to create a "release" for it.  You specify the branch/revision, give it a version number, and then github provides a direct link to download a zip of that revision.

Quote:it would also be helpful to tell people what Aerophones, Chordophones, Electrophones, Idiophones and Membranophones are

Yeah, I had to look those up!

---

Quote:Starting on E6 (C3=60) and up, there are no more releases on the piano and all keys should be given a long release of something like 10.0.

You don't really mean 10 seconds do you?  I just tried this, but it sounds like the key is never released.  The damper should silence the string in a fraction of a second.

Quote:Attack of 0.1 on soprano recorder sounds unrealistic (the instrument is supposed to have that chiffy start), so I removed it.

The reason I put that on is that a few notes have pronounced squeaks at the start that make them stand out from the rest of the instrument.  But I can move it to just those notes instead of the whole instrument.

Ah cool, I think I saw something about that releases feature. Will try it out once we think we're ready to go "prime-time" with what we have so far.

Yes, on pianos from either E6 or F6 (rarely sometimes up to G6) up there are no dampers and the notes sustain forever. Try it on a real one, this isn't a joke. 

http://art-facts.blogspot.com/2009/07/ho...-have.html

As pointed out, it's actually even used in many compositions intentionally.

Regarding the soprano recorder, I think putting them on the notes is the best method, but I can also edit the samples and try to remove the squeak without removing the chiff. I will probably redo the recorder samples at some point anyway, but these ones are already good enough that I think they do the job just fine.