Before buying the XSample Vol.6 (Woodwinds) I was fascinated by the idea of having chromatic sampled instruments (= each note sampled).
Today I think different: Having a unity note - an excellent sample - in the middle of three notes can give you very smooth results and saves a lot of speed and memory. It could be optimized by adding extra samples for extreme hi/lo notes.
The problem with chromatic samples is that you get a scale of many different and very indiviual samples that do not easily blend together to form a smooth instrument.
In fact an oboe playing a melody legato \"warms up\" while playing and so the whole instrument \"sings\" the tune as the notes blend together.
I found that this effect comes more easily with carefully interpolated samples rather than with the many chromatic and very individualistic sounding samples that had been recorded isolated.
With the XSample Oboe I had to cut out a number of strange sounding or badly played samples anyway and finally found the result better with some interpolation.
Also the speed/memory saving effect is very significant.
I vote for long, excellent sample and 2-3 notes per sample. Of course this is not valid for pianos etc., where chromatic sampling add a lot of realism.
Horst I\'m glad that you share my opinion. I done loads of sampling and I\'ve found that (especially for brass and woodwind) interpolation of samples is often a much better solution than chromatic sampled instruments. Especially with fast runs. When you play a fast run with an unbalanced chromatic sampled instrument, some notes tend to stick out and you get unconsistency in your phrasing. This is not so much the case with interpolated instruments because you get a more homogenous sound which aids to the consistency of your music. Future library developers should keep this in mind. Great post Holst!
I also feel that not every note need be sampled. I personally would like to see a feature called legato layer that the Ensoniq ASR10 had. It layered an attack layer with a non attack layer that was only heard in legato phrasing. It worked remarkably well on wind instruments. Just my 2 cents.
interesting ideas, but I definitely don\'t share them. (It\'s not the reason, that I\'m afraid, we have done a lot of work for nothing.) Here my reasons:
- Maybe there can be a single case, in which you would prefer a pitched tone. But watch out the whole instrument. It is loosing the character, because the very important formants are gone. For example the oboe has the main formants at about 1100 Hz, 2900 Hz and 4700 Hz. Going a semitone down (multiplying with the 12. root of 2) you are in the region of the Oboe d\'amore, being more extreme maybe a third tone, than you are listing to an \"english horn\" or oboe da caccia(with strange noise). Anyway if it sounds something like natural, you are definitely loosing the differences between very similar instruments, that can give the kick in your arrangement.
- For Emu and Akai we programmed this kind of sets as a saving memory preset. Every time when I play this sets, I think, I\'m playing a plastic instrument, specially if you are playing scales.
- I think the variety of notes in a sampler is small enough for alive sounds. I\'m glad for every variation.
- The feeling between chromatic setted sounds is much better than 3 equal sounds and than a change. You feel the changes much more.
- I >know< (if I\'m allowed to say that), it\'s the exception, if it really fits. With the most instruments this doesn\'t work. Try it with our new string collection (which is now in copy) or maybe with any other not pitched strings. Sorry, but that really sounds very cruel. You will, see this idea isn\'t the solution.
- The loading time is OK I think. (You are really loading one preset at time?). I often use also just one dynamic range from our mono sets (single presets at the end of the gig list) they are smaller and speed /memory are very OK and not bad for a bigger arrangement.
But finally it\'s up to you finalize the instrument for your taste. If it\'s your taste, it\'s very quick done in the Editor, killing the regions and drawing them in the gaps. But the other way is a lot of work. Sorry, but we will going on the >chromatic< way. But if there are many customers liking this kind of presets, we would think about making some articulation-files in our web.
Transposing will keep an (sometimes very nice) effect, but no natural sounding solution in our library.
The discussion about the astonishing new program Melodyne on the \"Million Dollar Idea\" topic is also germane to this topic. Melodyne--which is a very expensive professional program from Germany--has developed a way to pitch shift audio samples over a wide pitch range with little degradation of musical quality. It accomplishes this with balanced ratios among all of a samples parameters. You change the pitch, you are also changing the vibrato rate, the timbre, the attack, the volume, etc. accordingly. With Melodyne, it would probably be relatively easy to make perfectly even 12-note scales of samples.
The Melodyne idea might work if your instrument\'s C3 timbre was mathematically related to C4, as some instruments are, but it won\'t work if the instrument has special resonant or dead areas. How would the programme know this? It just interpolates everything from the original sample, while keeping the formant constant.
Imagine a singer. You sample a certain note which is in a comfortable part of his range, then \'Melodyne\' it up an octave and a half.
Now compare this version to the real singer performing an octave and a half higher.
They won\'t even sound like the same guy, because the Melodyne is too good. It has no problems going up an octave and a half, whereas the real singer\'s throat may undergo great strain doing the same thing, causing the timbre to be quite different.
As I said, this isn\'t the fault of the Melodyne programme, it\'s a natural fault of the instrument (or voice). A non-linearity which is always difficult for a comupter programme to take into account.
All we\'re talking about is cutting out bad sounding samples (regions) and extending the samples on either side of the gap we just created to fill it in. It only works when most of your samples are consistent and you don\'t have a single sample covering too large a range.
Again, I agree with Georg. If the chromatic sampling is good, you don\'t need to spread samples over wider ranges. It\'s only when the original sampling technique is poor, or the performance is inconsistent, that you need to cover up the bad patches with their better neighbours.
I find the woodwinds quite realistic in the Xsample library although a legato mode would definitely help for solos. All instruments have notes around the crossover points that may sound like a bad recording but actually show the true sound of that note. I play many woodwind instruments (clarinet, oboe, flute) and they all have these odd sounding notes in real life.