Comparing GOS, Miroslav and VSL

[Peet78]

Hi,

I'm using Garritan Orchestral Strings and the Miroslav Vitous samples. I keep on hearing great stories about Vienna Symphonic Library, VSL. I was wondering if anybody on this forum works with this package and if he/she could explain me the differences between the three.
I'm particularly interested in the performance tools which come with VSL. I know the Legato function and alternation functions in GOS, but I was wondering how they work in VSL and options they give you. Are they more than in GOS?

Regards, Peter

[gugliel]

I have used Seidleczek Advanced Orchestra (AO), GOS, and VSL (solo strings only), and recently EWQLSO Gold.

My experience is only based on the VSL horizons solo strings, but i think the other VSL legato instruments all work the same way. The 'performance tool' adds a key switch and a transposition to notes in a legato passage used with a legato instrument. Each 'legato' instrument note is really 24 keyswitched samples: the note with normal attack (start of a legato passage); the note with a minor second step up, major second step up, minor third step up, etc, up to an octave leap, then with a minor second step down, major second step down, minor third step down, etc, again up to an octave. In each case, the sample for a note includes a 'release trail' of the previous note at the appropriate pitch difference.

The GOS legato is more like a tiny bit of keyboard pedal applied (a 'release trail' of the same pitch as the previous note).

The VSL legato sound is better than GOS legato (imo) at the expense of large size and a little extra complexity. There is no special feature for legato in AO though there are some keyswitches called 'legato'.

The legato instruments end up being quite large in vsl, even with just one velocity level. I made a 'custom' version omitting the samples for skips larger than a fifth in order to fit a good selection of vsl instruments into a 1.5 gig pc.

You can do VSL legato and GOS legato in midi directly, omitting the 'performance tool' and the 'maestro tool'. Using the CAL language in sonar, for example, one can quickly (but not during playback) apply the legato keyswitches and transpositions to a midi track.

VSL's biggest failing is inconsistancy in timbre from one dynamic level to another, and it has a few minor tuning problems; GOS has some tuning problems too, some samples are simply out of tune with themselves and others waver in pitch or not exactly tuned to equal temperament; it is more consistant from velocity layer to the next layer. AO has many tuning problems in terms of wavering and exact tuning, but no samples that I recall out of tune with themselves, and not many velocity timbre problems (not many velocity layers, period). EWQLSO has tuning problems in the strings, too, both varieties, samples out of tune with themselves and samples not tuned quite correctly. Don't know yet about velocity/timbre inconsistancies. No special legato technique in Gold, and the heavy reverberation causes faster passages to blur badly (but I'm still new to using Gold, so this could be user error!)

Hope this helps. I've been a heavy strings user for a few years with different packages, and work with AO, GOS, and VSL daily.

[Dietz]

There are no obvious tuning-errors within our library, and even if you suspect one sample to be faulty, you are kindly invited to contact our support and/or to come to our large forum-site on vsl.co.at/forum . We try to fix any issue within hours, usually. -

The same is true for new customers, of course :-] ... Peet78 - please come to our site and listen to those hundreds of demos on vsl.co.at/demos to get the idea. Enjoy!

All the best,

[gugliel]

There ARE tuning errors in VSL (solo strings, anyway). They are not large, I don't mean to imply that; however the response in the VSL forums to my comments on tuning particular notes was dismissive, so the above statement is wrong. Don't mean to be hard on Dietz, but giving a blanket statement like the above annoys me. And the VSL patches are all programmed with pitch bend set to ZERO, so it is impossible to fix tuning without re-programming every instrument.

also, it is especially appropriate on this general forum to compare and contrast differing packages, so for a representative of the company to come and attempt to flatly contradict a user's point of view is bad. And THAT is the other fault with the VSL folks, perhaps American-style customer service doesn't make sense to them -- on their forums (happens here, too, occasionally) people get censored, cut off from access from time to time.

[David_Carter]

Peter - best thing to do is go to the VSL site and listen to the demos and watch the recent videos they've put up about the performance tool - about 15 minutes worth - they're very informative.

[gugliel]

Here is a quick example of both tuning errors and mis-matched timbre, from the vsl solo strings package, the half-second duration detache with four velocity layers and an up and a down bow (or rather, a 'bowing 1' and a 'bowing 2'): www.guglielmomusic.com/vsl_solo_example.mp3 Taken from the boring routine I have to use to run through a tuner (I use G-Tune), record the adjustments, write a Sonar CAL program to adjust, and then use in compositions. For adjusting timbre, I haven't yet found a way to automate it or even tabulate it -- requires a fair amount of touch up by listening and changing.

Shows both that the samples are not perfect and that they are pretty good anyway.

[dcoscina]

wow, that violin sounds INCREDIBLE! I don't really hear a big difference. Perhaps when you switch from the soft bow to a harder down bow, the note is a bit sharp but I attribute this to the vibrato in the playing style and not the programming.

If this is the worst VSL gets, man, I'm going to have to save my pennies for this library. Really awesome solo violin.

[Hans Adamson]

Gugliel,

I was a violinist up to 20. Not at professional level, but still. I don't hear any mismatched timbre. You have dynamic levels with the timbre that naturally corresponds to the applied attack/ bow weight. Where's the mismatch?

Hans

[janila]

I've played violin my whole life and must say that never ever is a violin player going to play exactly in tune, maybe a single note every now and then. Every note is somewhat out of tune depending from the events before and after the note. That is part of the violin sound. A poor player is random in his tuning, a good player is where he intends to be but rarely in the precise tuning you can create with your samples and tuners. Therefore you should tune every single note as a violin player would do concidering that you want to take that road. Still there are bigger problems when sampling solo string instruments and atleast I don't mind about a note being couple cents sharp, a sample being exactly in tune might not be any better in the musical context.

I didn't quite get the order of the samples in your demonstration. Which ones are different dynamic layers from the same program? Dynamic layers and crossfade programs are a big problem in most of the orchestral libraries and are especially difficult to do well in solo string programs as crossfading to different phases of the vibrato will always sound awkward. The same thing causes problems in the transitions between notes as the vibrato is a somewhat independent part of the violin technique that usually doesn't react to the note changes.

I'm also a bit annoyed about samples being in slightly different places in the stereo field. When listening with earphones the minor deviations make a solo line waver back and forth in the stereo field. The same problem exists atleast in VSL Opus1 and EWQLSO Silver. Have you noticed it? It doesn't matter in full orchestral texture but is fairly noticeable in solo parts.

[Marcussen]

peet - all in all I would say that VSL strings with say GigaPulse are the best out there and I own GOS, SISS and VSL

And the Solo Strings from VSL are indeed - amazing

[vinney57]

Well, Gugliel, er... having listened to the 'music' on your site I think your quest to expose a tuning error that nobody else can hear is a little ironic... :confused:

[FredProgGH]

I've played violin my whole life and must say that never ever is a violin player going to play exactly in tune, maybe a single note every now and then. Every note is somewhat out of tune depending from the events before and after the note. That is part of the violin sound.

Agreed. I spend half my time trying to make string libraries MORE out of tune. Not much- just subtle changes in intonation.

[gugliel]

Well, thanks for all the comments. In particular, some of you are right that real players never play precisely in tune. The problem is that a real player, a good one anyway, plays out of tune to make the line and/or harmony MORE musical. Just using an out of tune sample doesn't achieve this -- would have thought that was obvious! So to achieve this in midi, one must first identify the out-of-tuneness in the samples, then identify the needed corrections in the music, then apply a double correction to your music. If you hear out-of-tuneness in my music examples (and they are still there, to be sure), it's from four reasons, one good imo and three bad: (1) I'm trying to write modern music that benefits from the tuning inflections real players provide (2) all the samples I use are at least a little out of tune in unpredictable ways (3) some of my corrections are automated and make automated mistakes I haven't yet found and corrected (4) some of the recordings are from my pre-tuning days and combine different sample sets (AO, GOS, VOTA) that are pretty far out of tune with each other -- leading to today's quest.

[edit] and I'm like any car buyer ... once I've bought something, I want to talk about it, criticise or praise it, and still recommend it to my friends. VSL Horizons solo strings is a fine package, worth almost the great amount they charge for it! [\edit]

[gugliel]

One more thing -- the problem with mismatched dynamic levels is that a key velocity change of only a couple of ticks changes the sounds from one level to a sharply different level, sometimes in a different panning, as janila said. My example used pretty dramatically different dynamics to separate the sounds for tuning recordation.

To get around this problem, vsl solo package offers programs with which one might switch timbre by means of the mod wheel; one must switch bows, too. I modified these to switch timbre by a combination of a midi controller and key velocity and switch bows by means of the mod wheel.

[Bruce A. Richardson]

I would say that the one tip I'd offer when using VSL (or any 3-4 layer sample) is that you want to stay within a particular velocity-split in most phrases. For example, if the overriding dynamic level of a phrase if mezzo-forte, then I'll play in the phrase, and edit any note's velocity which "breaks through" the intended timbre. I then use an expression envelope (in SONAR) to help with any shaping the phrase needs.

Until there are 32-layer string libraries, I don't see any other way to get timbral consistency. Of course there are musical exceptions where you'd want to get a subito timbral shift. But that is not true in most cases...most melodic phrases want to express themselves within a relatively stable timbral range without going too "over the top."

I refer to a 32-layer library somewhat in jest, obviously. This is just one of those difficult points of sampling. The more layers you have, the more opportunity for inconsistency. I'd say the Dan Dean libraries, at eight layers, are good examples--not of the inconsistency, because they are very consistent laterally across notes at a given velocity. But even with eight layers, you still have the jumping issues if you're not careful, and ultimately, I think even with layers and layers and layers, you'd always have notes jumping around timbrally which need adjustment. It just comes with the territory.

By the way, someone mentioned notes jumping around in the stereo field. This, too, is one of those devils of sampling which is almost impossible to control. Instruments are highly directional, even seemingly non-directional ones. They all set up a phase relationship with the room and the microphones, and even a tiny movement by the player can send a completely different wave pattern into the air than the previous one. This happens in all recordings. The reason it's so exposed in sampled recordings is that the recording isn't continuous, so we don't hear the actual movement. We hear only the very granular, out of context, result of movement. Unless you use instrument stands (some guitar libraries do) or devise very elaborate methodologies for the players (which they will not always remember to follow) it is difficult to make samples "stand still." Sample developers are becoming more sophisticated at this all the time, but it will always be difficult to achieve an absolute stability of imaging, because in reality, that doesn't exist in any recording...it's just that without the note-bound granularity, you'd never hear it.

[Eric G]

I would say that the one tip I'd offer when using VSL (or any 3-4 layer sample) is that you want to stay within a particular velocity-split in most phrases. For example, if the overriding dynamic level of a phrase if mezzo-forte, then I'll play in the phrase, and edit any note's velocity which "breaks through" the intended timbre. I then use an expression envelope (in SONAR) to help with any shaping the phrase needs.


That's exactly why I wish more orchestral libraries followed the GPO performance paradigm. When playing string or wind samples (either solo or section), doesn't it make much more intuitive sense to control attack strength, rather than dynamic level, with velocity, and use the mod wheel to "blow" or "bow" through a phrase? If this were more consistently followed, velocity-induced timbre spasms wouldn't be an issue.

My ideal programming: polyphonic programming, switchable to VSL-style legato with the pedal, with modwheel crossfades (or timbre morphs like the old e-mu filters!) through however many dynamic levels necessary, and keyswitches for any other articulations that this scheme couldn't cover convincingly. How many dimensions would that take?

[Dietz]

There ARE tuning errors in VSL (solo strings, anyway). They are not large, I don't mean to imply that; however the response in the VSL forums to my comments on tuning particular notes was dismissive, so the above statement is wrong. Don't mean to be hard on Dietz, but giving a blanket statement like the above annoys me. And the VSL patches are all programmed with pitch bend set to ZERO, so it is impossible to fix tuning without re-programming every instrument.

also, it is especially appropriate on this general forum to compare and contrast differing packages, so for a representative of the company to come and attempt to flatly contradict a user's point of view is bad. And THAT is the other fault with the VSL folks, perhaps American-style customer service doesn't make sense to them -- on their forums (happens here, too, occasionally) people get censored, cut off from access from time to time.
Of course comparisons are ok on these pages. And almost everywhere else, too! :-]

Nevertheless I have to reply to your message, as you refer to me as a "representative": I didn't mean to "flatly contradict a user's point of view", I kindly invited you (and everybody else, as you are already a frequent guest) to come over to our site to make use of our extensive support options - IN CASE that there is something wrong with one of our products. So please don't reduce it to an American/European cultural thing, we try to think and act globally.

BTW - my messages here reflect my own private opinions; I identify myself as part of the Vienna Symphonic Library so clearly for the only reason to avoid the smell of under-cover propaganda.

And a final note: you know exactly _who_ was banned for three months, and why. (Most certainly not for a healthy comparison between (seemingly) competing products.) Moderation of a forum is just another way of making it a pleasant place to stay for the other 99.999% of our users (for you, too!).

Ok ... back to samples, sounds, and tuning issues! :-]

PS - thanks, Bruce, for explaining the principle behind those typical movements of recordings in the stereo field. You're right on the point.

[Beckers]

I'm also a bit annoyed about samples being in slightly different places in the stereo field. When listening with earphones the minor deviations make a solo line waver back and forth in the stereo field. The same problem exists atleast in VSL Opus1 and EWQLSO Silver. Have you noticed it? It doesn't matter in full orchestral texture but is fairly noticeable in solo parts.

Just to add to what Bruce said, much of this wandering in the stereo field is normal, player movement apart. String instruments are by nature highly directional, emitting different freqencies preferentially in certain directions (walk around a violinist and hear the timbral change. The change itself varies with pitch; at some pitch points this is sudden, over a semitone). Unlike player movement, this is predictable and a constant. The closer mic'd, the more pronounced it is. (This is also the reason why you cannot capture the full timbre of a violin with a single set of XY stereo mics close up..)

A live violinist playing solo tends to move a lot, but this is progressive from note to note, therefore not so noticable. This movement also masks the perceived source movement caused by the natural directionality of the instrument. If you nail a player to the chair during a sampling session you still get changing directionality. In music recorded with these samples this is then not masked by progressive note to note body movement, and may therefore be seem unnatural, even though it isn't.

One answer is to only use mono samples. I hate mono samples, for specific reasons. In fact I've spent some time in "repositioning" my own samples with various stereo controllers with fair success. -Simple panning doesn't work, partly because it does not affect the timing. (Correcting the timing by splitting the stereo channels, with panning, works better but causes other problems, and, you know, I also have a life)

The thing about samples is that they have to be perfect. Notes played by a live player always have flaws, tuning, scratches, breaks, inconsistencies, but these are all one-offs. On sample play back, any repeat of identical flaws will stand out like a ton of bricks.

That's why you need perfect samples, or many alternative samples. Deliberate imperfections can be performed in midi.

[Dietz]

Thanks for a concise and insightful explaination. Would you mind if I link to this message if someone asks again ...?

All the best,

[janila]

The change itself varies with pitch; at some pitch points this is sudden, over a semitone). Unlike player movement, this is predictable and a constant. The closer mic'd, the more pronounced it is. (This is also the reason why you cannot capture the full timbre of a violin with a single set of XY stereo mics close up..)It is nice to see that people bother to write concise answers. I want to point out that I don't concider this a major problem but slightly annoying in some occasions. I'm continuing this discussion just because I'm interested, not because I want to complain about the panning variations.

I obviously have to clarify my point. I wasn't wondering where the stereo variation come from but asking indirectly that are these variations edited after the recording? The theory is certainly simple as the phase difference between the two microphones makes the audible panning differences between the samples, moving the instrument a couple of inches isn't audible in mono setup where only the level is a factor. Therefore all there is to do is to make a tool that does the panning by moving the right channel earlier when panning to the right and moving the left channel earlier when panning to the left. This process is clearly completely nondestructive and doesn't affect the sound quality of the sample.

I understand that there are tons of samples in any orchestral library. Still these panning problems are noticeable only in some of the samples in some of the instruments and there is no need to edit them all. Is there any kind of compensation done for this or are the samples "only" cut to the appropriate length, noise reduced and tuned before they are mapped to the actual programs? Just out of curiocity, can you Dietz inform us about the tuning, how often the samples are retuned in the cutting room and how often they just "happen" to be right? What tools are used for the tuning?


In fact I've spent some time in "repositioning" my own samples with various stereo controllers with fair success. -Simple panning doesn't work, partly because it does not affect the timing. (Correcting the timing by splitting the stereo channels, with panning, works better but causes other problems, and, you know, I also have a life)This is why it would seem logical to do the editing in the sample library production company as it would have to be done only once instead of many end users doing it by themselves. Just a thought, nothing major. ;)

[Beckers]

The theory is certainly simple as the phase difference between the two microphones makes the audible panning differences between the samples, moving the instrument a couple of inches isn't audible in mono setup where only the level is a factor. Therefore all there is to do is to make a tool that does the panning by moving the right channel earlier when panning to the right and moving the left channel earlier when panning to the left. This process is clearly completely nondestructive and doesn't affect the sound quality of the sample.



It depends how the samples were recorded. If a single XY setup was used, say for a solo instrument, there would be no timing or phase difference if the player moved (the mouths of the mic's are placed together at 90deg). So all you've got is amplitude difference. But this is complex, because the side signals are affected much less than the mid. Simple panning, i.e. tweaking the volume of one channel, will affect the M and S signals equally and you get a change in the sound space as well as shifting the position of the instrument. There is software that can shift the central signal independantly of the side signal. The cheapest is PSP Stereopack.

I recorded my own violin samples a couple of years back. I used a carefully measured multi mic setup, which captured the violin timbral spectrum beautifully, but despite bolting the player to solid masonary she managed to move (they just must), and I ended up with timing problems (retrospectively predictable). I tried to correct this on the worst samples, by splitting the channels and moving them. The diffence was mostly 8 - 20 samples, at a samplerate of 44.1K, that's -what, 0.5ms, but enough to cause a significant shift. However shifting the timing of one channel caused combfiltering. (Because of the carefully measured mic positions (I guess), my samples were otherwise mono-compatable i.e. no combfiltering: even when the player moved, causing a stereo field shift, -> go figure).

I left the timing well alone, but a quick tweak of several parameters with a plug-in stereo controller made the whole thing tolerable.

Every sample needs tuning. I used Melodyne, which hate but couldn't do without.

[Dietz]

[...] Just out of curiocity, can you Dietz inform us about the tuning, how often the samples are retuned in the cutting room and how often they just "happen" to be right? What tools are used for the tuning?
[...]
Our recordings are not "re-tuned" in that sense that a recorded "C" is used as a "C#". What gets corrected most of the time are otherwise immaculate recordings that have too much "human touch", or recordings from the extreme ranges of an instrument. - Every single sample runs through our production cycle, going through the hands of up to eight or ten highly speczialized editing engineers, that's why it's hard to tell how many samples are tuned and which ones go out as they came in; but as every sample gets checked for tuning (and other aspects), it is very likely that it is corrected to a small amount, too.

The tools range from the built-in features of Sequioa (our main platform), to adapted versions of Melodyne and some proprietary solutions.

HTH.

[Bruce A. Richardson]

Regarding the corrective 'repanning" of sample positions...

Unless the library was made by a total hack (not likely these days), the microphones never move during a session. The room's "air" is always being captured in the same position, the "bounce" of first-generation reflections always coming off the same non-moving walls.

So, if you repan the individual notes to put the "instrument" in exactly the same place, you are panning the "walls" into a different position.

This cure is actually worse than the problem it proposes to solve, because there is much more imaging information hitting the microphones from the wall bounce than from the original source.

In VSL it is really not that bad. In fact, that is one of the most impressive things about VSL, that they were so bold in recording very loud and hard first-generation reflections, using very vulnerable mic plots, yet the imaging is remarkably solid. Knowing how hard this is to achieve, it probably points to literally thousands of samples that died on the cutting room floor, and many, many takes to get instruments as consistent as these--across so many articulations.

[Peet78]

Hi guys and girls,

Thanks for your enthousiastic replies!!!
The reason I posted this thread was that I don't seem to get a nice string sound. I'm learning to use mainly GOS, and I find the violins LEG EXP to 'dead' or synthesizer-like. I like the Grand Sustain Vlns a lot, but they have the little dip in every attack so if you want to play a nice Legato line, it sounds a bit strange.
I've started to layer different sounds in GOS, for example the Grand Sustain violins for their beautiful vibrato and the VLN EXP LEG instruments to get a line more Legato. But I have the idea the sound gets thicker then.
I'm really interested in how you guys get a nice string sound and what you do to achieve this. Do you layer too? Or do you think I should make this a new thread???
Also, I'm experimenting with the MOD wheel which in GOS is for crossfading 4 dynamic levels, but to me the changes sound to harsh. I liked the idea of one of the repliers to use the MOD wheel only to change colour and use the Volume control to master the dynamics. E.g. I take the softest velocity level and do the dyn with volume, but then it's hard to get a louder sound. How does this sound, what do you people think is best???
And what do you think about post-editing. For example if I post edit the Grand Sustain Violins, I can get it quite Legato (I edit the length in the Matrix editor in Logic 5).
Also, is there a way to combine Miroslav Samples with GOS, as they seem to be in a different spacial perspective.
Well, as you hear, I've still got a lot to learn. Hope you can help me!!!

Have a nice day!!

Regards, Peter

[Robert Kooijman]

"So, if you repan the individual notes to put the "instrument" in exactly the same place, you are panning the "walls" into a different position.
"
Great observation from Bruce. Also good to see that these issues are taken more seriously now in the sampling world. We slowly get there...:)

[Beckers]

So, if you repan the individual notes to put the "instrument" in exactly the same place, you are panning the "walls" into a different position.

This cure is actually worse than the problem it proposes to solve, because there is much more imaging information hitting the microphones from the wall bounce than from the original source.



Except, as I pointed out in a post above (no doubt unread), that with a stereo controller it is possible to improve this. This compares signals and separates initial waves from their delayed counterparts. This allows you to manipulate the direct (mid) signals separately from the reflected waves (side signal). So you can move the centre without changing the space. The closer the mic'ing the tbetter this works. In a mix any residual changes to the space in a sample is unnoticable.


You can also collapse or expand the space. This can be useful to match diffent hall sounds in different libraries, although it is a blunt instrument, and doesn't work as well with far mic'd samples.

Also, is there a way to combine Miroslav Samples with GOS, as they seem to be in a different spacial perspective.
Regards, Peter
As above, together with matching the releases and EQ.

[Beckers]

Hi guys and girls,

Thanks for your enthousiastic replies!!!
The reason I posted this thread was that I don't seem to get a nice string sound. I'm learning to use mainly GOS, and I find the violins LEG EXP to 'dead' or synthesizer-like. I like the Grand Sustain Vlns a lot, but they have the little dip in every attack so if you want to play a nice Legato line, it sounds a bit strange.
I've started to layer different sounds in GOS, for example the Grand Sustain violins for their beautiful vibrato and the VLN EXP LEG instruments to get a line more Legato. But I have the idea the sound gets thicker then.
I'm really interested in how you guys get a nice string sound and what you do to achieve this. Do you layer too? Or do you think I should make this a new thread???
Also, I'm experimenting with the MOD wheel which in GOS is for crossfading 4 dynamic levels, but to me the changes sound to harsh. I liked the idea of one of the repliers to use the MOD wheel only to change colour and use the Volume control to master the dynamics. E.g. I take the softest velocity level and do the dyn with volume, but then it's hard to get a louder sound. How does this sound, what do you people think is best???
And what do you think about post-editing. For example if I post edit the Grand Sustain Violins, I can get it quite Legato (I edit the length in the Matrix editor in Logic 5).
Also, is there a way to combine Miroslav Samples with GOS, as they seem to be in a different spacial perspective.
Well, as you hear, I've still got a lot to learn. Hope you can help me!!!

Have a nice day!!

Regards, Peter

I also struggled with GOS to begin with, particularly its timbre. The "dead"ness is worse on the EXP crossfades, because a lot of the time 2 samples are playing, which in effect doubles the number of violins playing and further "irons out" the vibrato. However riding the dynamics with volume control rather than the MW EXP crossfades sounds even more synthy.

Every string library has its own timbre. I guess it's partly to do with mic distance and hall etc. But I've often wondered about GOS. The sections recorded by Garritan were a collection of master violins, including Strads, Guarneris and Gaglianos. Master violins have a strong bridge resonance at 2000-2500. This is great for solo playing, as it cuts through the orchestra, like an opera singer's formant. But even more noticable is the continued high responce through to 4000Hz and beyond. (Factory made violins show a stronger response around 1000Hz i.e. more nasal).

There is no string orchestra in the world that comprises this sort of collection and we're not used to hearing this. I reckon this contributes to the unique formant (frequency response profile) of GOS and makes it sound unnatural.

I also notice that the timbre change across the dynamics is remarkably consistent up the pitch scale. In particular it sounds like the fundamental harmonic diminshes consistently with samples of decreasing velocity. This sort of consistency is hard to achieve by players (timbre depends a lot more on bow pressure than velocity) and I wonder whether the samples were EQd this way afterwards. (I could be wrong I could be wrong. My experience is limited to solo and small sections, and perhaps inconsistency completely evens out with large sections. Or else the samples could have been selected from huge numbers.) Even small cuts in the fundamental harmonics would make these master violins sound crisp indeed.


With a little careful EQ on a rendered GOS track you can change the GOS formant to resemble a "regular" section more closely. Avoid touching the primary and secondary bridge modes around 2000 and 4000Hz respectively, but cut with a narrow Q in between, at 3000Hz. Also shelve off after 5000Hz or so. Avoid touching much below 1500 with narrow Qs. But cut at 1500, enough to remove some of the "ahng" colour.

You could make GOS sound warmer by boosting the fundamental harmonics a little. (I don't like the "Warm" versions in GOS; the LP filters cut into the bridge resonance modes and makes them sound less "violinic"). You can't do that on a rendered track because the fundamental frequency of one pitch may be occupied by other harmonics from other pitches. You'd have to export the samples and EQ at the right frequency of each sample. OR you could ... render the GOS part as seperate octaves by cutting and pasting over several tracks in midi. For each track you'd then have a known range of frequencies where there are only fundamental harmonics. (I did this routinely when I first got GOS 3yrs ago, but then found other ways). BTW GOS is much more pliable than MV.

Edit:


'dead'

Oh, yes; to revive, use the EXP versions and layer in at least two solo violins with vibrato, played seperately onto their own midi tracks (or copy the midi track for GOS and make changes), EQ matched, only just audible. Got to go, but if you want it to sound more intimate I'll explain a way to do that.

[Jeannot Welter]

What?!
A section consisting of Strads, Guarneris, Gaglianos "sounds unnatural"?
Would you rather have a section of $300.00 violins made in China, out of premium plastic with all steel strings?
Having all these instruments in one room to record samples is never going to happen again. Gary Garritan made history with this collection.
Having heard GOS2, I can assure you that it will be a pleasant surprise.

Jeannot Welter.

[Beckers]

What?!
A section consisting of Strads, Guarneris, Gaglianos "sounds unnatural"?
Would you rather have a section of $300.00 violins made in China, out of premium plastic with all steel strings?
Jeannot Welter.

Jeannot, how often have you conducted or heard a section consisting purely of these sort of instruments? My point is that we're not used to hearing whole sections of $3,000,000 violins, neither $300 violins, but more like $3,000 instruments. If you agree a Strad sounds different to a $3,000 instrument, then you must agree GOS is bound to sound different to a regular section, which may by many ears be interpreted as "unnatural".




Having all these instruments in one room to record samples is never going to happen again. Gary Garritan made history with this collection.
Jeannot Welter.
Kinda makes my point doesn't it?

Can't wait to hear GOS 2. (Were the samples EQd or FFT filtered in any way?...)

[Jeannot Welter]

I have never conducted a string section consisting of only 3.000.000.00 instruments. However I had the privilege to listen to artists like Oistrakh, Ricci, Amoyal, Chung, Heifetz ( Daniel that is ) etc. from less than a few feet away, most often conducting while they were performing as soloists.
I can attest that these artists' instruments sounded very "natural".
The most striking aspect of their performances was that when they were playing in concert they sounded different from when they were recording.
In concert, in halls filled with 2,000+ people they never really produced a normal p or pp, knowing that they had to reach the last listener in the last row in a very big hall.
When we recorded they fully exploited the whole dynamic range of their instruments, knowing that the microphones would capture all the nuances of their performance.
If you ever listened to the Alban Berg quartet in concert and then listen to their recordings, you will notice the same difference.
When Gary recorded these instruments at Lincoln Hall, there was no need for these players to produce maximum volume from their instruments because there was no public, thus the frequency response behaviour that you describe did not occur.
In what GOS1 is concerned, there have been many suggestions about eqs, some eqs more mortal than others.
I found myself guilty of trying to eq GOS and in tandem with Ernest Cholakis spent half a year developing timbral impulses to make sampled strings sound more natural, organic ... whatever..
Although the results were stunning in blind AB comparisons, I still was not satisfied.
After years of pleading with Gary, he finally let me listen to some of the original recordings of GOS. I was absolutely "blown away" by what I heard.
No eq can improve the sound of these strings. On a side note the same is true for the solo strings that Gary recorded.
Now the legitimate question is: why don't these qualities fully shine through in GOS1?
GOS was recorded 5 or 6 years ago. Since then software used for processing samples has made leaps of progress. Applications developed by companies such as Samplitude/Sequoia, Algorithmix now allow pratically non-invasive processing of samples and the results are amazing.
I apologize for this lengthy post. But I think it also answers the question why GOS2 and Garritan solo strings were so close to release and have not been released yet.
Is GOS2 a string library that I could not live without? Absolutely.
I have endorsed GOS1 and I fully endorse GOS2, I just hate the time it takes to get it to the public, in part because of Gary's many other projects on which I won't comment.
I hope that I have not violated any NDA and that Gary is not going to kill me for this post...

Jeannot Welter.

[Beckers]

I have never conducted a string section consisting of only 3.000.000.00 instruments. However I had the privilege to listen to artists like Oistrakh, Ricci, Amoyal, Chung, Heifetz ( Daniel that is ) etc. from less than a few feet away, most often conducting while they were performing as soloists.
I can attest that these artists' instruments sounded very "natural".
The most striking aspect of their performances was that when they were playing in concert they sounded different from when they were recording.
In concert, in halls filled with 2,000+ people they never really produced a normal p or pp, knowing that they had to reach the last listener in the last row in a very big hall.
When we recorded they fully exploited the whole dynamic range of their instruments, knowing that the microphones would capture all the nuances of their performance.
If you ever listened to the Alban Berg quartet in concert and then listen to their recordings, you will notice the same difference.
When Gary recorded these instruments at Lincoln Hall, there was no need for these players to produce maximum volume from their instruments because there was no public, thus the frequency response behaviour that you describe did not occur.
In what GOS1 is concerned, there have been many suggestions about eqs, some eqs more mortal than others.
I found myself guilty of trying to eq GOS and in tandem with Ernest Cholakis spent half a year developing timbral impulses to make sampled strings sound more natural, organic ... whatever..
Although the results were stunning in blind AB comparisons, I still was not satisfied.
After years of pleading with Gary, he finally let me listen to some of the original recordings of GOS. I was absolutely "blown away" by what I heard.
No eq can improve the sound of these strings. On a side note the same is true for the solo strings that Gary recorded.
Now the legitimate question is: why don't these qualities fully shine through in GOS1?
GOS was recorded 5 or 6 years ago. Since then software used for processing samples has made leaps of progress. Applications developed by companies such as Samplitude/Sequoia, Algorithmix now allow pratically non-invasive processing of samples and the results are amazing.
I apologize for this lengthy post. But I think it also answers the question why GOS2 and Garritan solo strings were so close to release and have not been released yet.
Is GOS2 a string library that I could not live without? Absolutely.
I have endorsed GOS1 and I fully endorse GOS2, I just hate the time it takes to get it to the public, in part because of Gary's many other projects on which I won't comment.
I hope that I have not violated any NDA and that Gary is not going to kill me for this post...

Jeannot Welter.
You misunderstand me. Of course Strads and Guarneris sound natural. But you can pick them out of a lineup blindfolded. You would expect a section of them to sound different to a usual section.

The sounds when played back from a sampler never sound the same as the original sounds. I don't know fully why that is, but I think it is more to do with psycho-acoustics than with the recording equipment or the processing and sampling software (though I agree recent libraries have been closer).

The fact is that, additionally, GOS does have its own sound, and I was musing on why.



And I would be surprised if Gary wanted to kill you for making people salivate in anticipation. (But he might want to kill me for what I said, after he's shelled out for all them fiddles)

[Jeannot Welter]

"The sounds when played back from a sampler never sound the same as the original sounds. I don't know fully why that is. I can't believe it's the recording equipment or the processing and sampling software (though I agree recent libraries have been closer). "

The audio engines in the different applications are of variable quality. A sample played back in GS sounds different from the same sample played back in Kontakt or Halion.
Kind of frustrating for developers.

Jeannot Welter.

[FredProgGH]

Since we have someone with intimate knowledge of GOS 2 in our midst, is there anything you're at liberty to say about it?? Especially just how it differs from/improves on the first installment??

[Jeannot Welter]

This is difficult to put into words. Given the advances made in processing technology, the main statement that I can make is that GOS2 faithfully reflects the original recordings, letting the splendor of the instruments that Gary recorded shine through in an unobstructed fashion. This processing technology was not available 5 years ago.

Jeannot Welter.

[Beckers]

This is difficult to put into words. Given the advances made in processing technology, the main statement that I can make is that GOS2 faithfully reflects the original recordings, letting the splendor of the instruments that Gary recorded shine through in an unobstructed fashion. This processing technology was not available 5 years ago.

Jeannot Welter.

So are you saying:

that (A) GOS1 sounds different to the original sounds purely because of a degradation due to the processing software originally used, which has now been sufficiently improved?

Or (B) were the samples originally audio edited for a specific reason, but resulted in a change in the way GOS sounded, which has been revised in GOS 2, using new technology?

Or (C) is the cause for the difference between the original recordings and GOS1 inadvertant/ not identified. But this has been compensated for in GOS 2 by further, new, processing?

Or (D) none of these.

Or (E) it's pointless asking because you do not want to be found beaten to death with a rolled up NDA.

[Jeannot Welter]

D and E.
Just wait until it is ready to go and you can judge by yourself.
I hate to describe sound quality with words, expectations can be very different.
However, Iam very impressed with what I am hearing.

Jeannot Welter.