I never have wanted to "record" or dub to a final master directly out of MIDI programming since when you do that, you lose the mix which allows another entire arsenal of tools to work with. You also make it necessary to have an enormous and flawlessly running computer system... [[:|]]
-
Wonderful topic Bill, and in direct relation to sample performance.
For those who started experimenting with samples many years ago, as did I, the quality often had us cursing, and spening large amounts of time re recording, and experimenting, trying to get that real sound. In a way, the 'roughness' added to the humanity of what we heard, and gave samples, in particular strings, a unique sound, not always true in pitch or quality of even sustained tone, but human nevertheless. And although the humanity was welcome, aiding the overall sound, more often than not, the uneveness of recording, or articulation, meant huge chunks of time taken with finessing the final product. And orchestral performances left a large gap in credibility between a sampled recording and one with live muso's playing a group from notes.
Now, finally, we have a consistent set of sampled sounds that we can rely on to be mp when we require, with consistent tone.
But is this enough? No. And it's not VSL that provide the limitation, on the contrary, they provide the quality and consistency that enables us to perform with confidence, knowing the relative values of volume, tone, and articulation are constant.
The thinness of strings isn't a recording error, but a perception of an instrument (s) from a perfomers POV. That is, we hear the strings as thin, becuase we hear 'live' with many inconsistencies, and variations, however subtle they may be. The action of drawing a bow across a string is not one of controlled friction within the context of a performance, but just one part. The player will emphasise, however slightly, the first beat of a bar, or cut that second staccato a little shorter because he's mentally preparing for that next stretch of hand position. And there are preconceptions of how a certain piece of music should be played, according to the conductors wishes, and performances from others. All of this affects the 'mental aural picture' the performer has, and he or she will play accordingly.
The first violin player may hear a fine section performance of the 1812, and decide the next time he plays he will lean more into the offbeat quavers in the descent before the finale, just because it sounds different.
Sustained notes are most difficult because (a) WE play them with too much consistency, (b) WE don't have set firmly enough in our minds the performance or tone and articulation we want, and (c) WE tend to play to the samples, instead of working to extract the performance we truly want. If you ask a violinist to play a top C several times, each time will be different. And the variation of tone, bow pressure, and the players own assumption of what you want will produce that which we seek, humanity, and a thicker sound, not lost in the mix. It's always been my practise for years, based on experience and the wisdom of others, to record violin sections not once, but two or three times, each time played live, and each sample section with it's own variation in pitch and bow sustain consistency.
And when i finally get to purchase VSL, my first task will be to construct violin sections from the original samples into, 1st Vi a, 1st violin b, 1st violin c. etc. Each with changes by a tick or two in pitch, and ech line played live, as my own variation in performance will give enough difference to create that human element.
(Assuming of course that i'm human!)
Regards,
Alex.
-
I thing William has set up a very important point. Thje violins are the worst and until now I have always avoid using violin sections. I tried to find an answer to the problem elsewhere (having perhaps a wrong hypothesis about the reason of the problem, but anyway):
http://www.soundsonline-forums.com/showthread.php?t=2241&highlight=string+section
Thank you also for others of your good points. LG
-
@Dietz said:
lgrohn, the superlative "worst" seems to indicate that everything is "bad" about our samples. Is this really the way you feel about our work ....? What a pity. :-/
I wouldn't take it to heart, lgrohn feels that way about all libraries, so you are not alone [8-)]
DG
-
No, no. Perhaps the proper word would have been "the most difficult", compared to horns etc. Sorry. As a user of EWQL and owner (but not user anymore) of Garritan, I still think that VSL still has the best demos.@Dietz said:
lgrohn, the superlative "worst" seems to indicate that everything is "bad" about our samples. Is this really the way you feel about our work ....? What a pity. :-/
I guess that this has happened to everybody:
When you have got one of these great new software it is so much better what one has used earlier ( I used Roland JV1010). But after some months of use one starts to get to know some limitations. I think critical people should be listened because that's one way for future development.
By the way, EWEQL just cancelled my writing rights on their forum. Garritan did it last spring. They just couldn't stand someone writing of the limitations of their products. Am I near of that on VSL Forum, yet?
-
@lgrohn said:
[...] By the way, EWEQL just cancelled my writing rights on their forum. Garritan did it last spring. They just couldn't stand someone writing of the limitations of their products. Am I near of that on VSL Forum, yet?
What makes you think so?
/Dietz - Vienna Symphonic Library -
@lgrohn said:
[...] By the way, EWEQL just cancelled my writing rights on their forum. Garritan did it last spring. They just couldn't stand someone writing of the limitations of their products. Am I near of that on VSL Forum, yet?
What makes you think so?Actually nothing. But I just always try to be honest, even on the risk of being cancelled. And of course, many times I am just wrong and would enjoy if I am corrected. I call it learning.
-
We all enjoy a vivid discussion here on these boards, and I think it's safe to say that we can stand an healthy amount of criticism. The only point when things got ugly was when the same wrong or meaningless statements were repeated over and over again, or personal attacks got out of hand.
So - don't hesitate to let us know what you think, but be prepared that we have _our_ strong opinions, too! [;)]
All the best,
/Dietz - Vienna Symphonic Library -
@lgrohn said:
[...] By the way, EWEQL just cancelled my writing rights on their forum. Garritan did it last spring. They just couldn't stand someone writing of the limitations of their products. Am I near of that on VSL Forum, yet?
What makes you think so?Actually nothing. But I just always try to be honest, even on the risk of being cancelled. And of course, many times I am just wrong and would enjoy if I am corrected. I call it learning.
Lgrohn... VSL is also the best comapny when it comes to its forums. They take far more critisism and positive comments about competing libraries than any other company. They have such enourmous class, and in the time i've been here I have only seen one or two bannings. Hell someone posting here is known to try obtain illigit versions of VSL (and posibly owning a few) and still has not been banned. So I think you can take it easy [:)]
-
" Thje violins are the worst and until now I have always avoid using violin sections. " - lgrohn
I changed the name of this thread because lgrohn really distorts what I was trying to discuss. The violins sound great out of the box. I am not talking about their sound in isolation, I AM TALKING ABOUT THEIR USE IN PIECES OF MUSIC.
This thread is not critical of any sample library (except that one another company sells that I didn't name) but rather of the use of violins or strings in general resulting in a thin sound.
I am trying to get ideas on the treatment of sampled strings, particularly violins, in both midi and mixing. The kind of thing Jay wrote is exactly what I am hoping people might want to discuss, not a bunch of negative trashing posts and bickering.
-
It is a very good subject, and jbacal and others have the beginning of the answer, perhaps: much wider use of dynamic, timbral, and pitch expressiveness. One difficulty I have been having is trying to 'espressivize' a line as a solo, when any live player would be responding to the whole mass of music into which the line might fit. I still don't think that simply eq'ing the whole of a string line makes sense, though modifying individual recorded notes as part of making them expressive might work (but getting us perilously close to the same total length of time involved as the 15 years of violin study)
-
Easy, easy, dear forum members - no hard feelings in this thread. [:)]
*****
William, the quest for "the" string sound is probably as old as orchestral recording itself. Leaving the more obvious main factors like the composition, the arrangement, the performance and the acoustics of the surrounding room aside for a moment, I dare to suggest that even in seemingly puristic productions the techniques of sculpturing this sound were always used to their fullest extent; the style of doing so was dependant on the respective era, of course.
Given the "raw" nature of our samples (in the best sense of the word), many ways of audio-processing make sense. This ranges from simple, static EQing to sophisticated DSP like convolution - the latter not only for reverberation, but also to mimic a certain signal path, or even "dry" resonances.
Personally, I tend to use a combination of several tools. Most of the time I start with some basic EQing, emphasizing the "body" of each section, while getting rid of the (mostly scale-dependant) resonances that colour the sound too much. The next step ist to tame the "shrillness" (is this a word?) between 2 and 4 kHz. This is _very_ much dependant on the arrangement: Sometimes it is necessary to cut here two or even three small frequency bands by 8 or even 12 dB; on other occasions, this would make the overall impression too dark and indirect, and a broad dip by 3 dB centered around 3.5 kHz is more than enough. More often than not, I add some kind of harmonic distortion, be it from analogue equipment (older tube gear comes to mind) or on behalf of saturation coming from tape-emulations or the like. This blurs the pureness of the harmonics in a good way and adds interest and variation to the sound.
In pieces with extreme dynamic changes from pp to ff it is sometimes impossible to find one setting for all occasions. In these cases, I like to use dynamic EQs (resp. multiband compressors) that reduce the volume of a certain frequency range only if a certain threshold is reached. This seems to be tricky, but the results are self-explanatory once you try it.
While it seems to be contradictory to what I just described, I sometimes _add_ some very high treble to violin-sections, for example a soft shelving EQ above 12 kHz or higher. Here, the quality of the EQ is crucial, of course - this is true for virtual orchestration as well as for live recordings.
The final step is some kind of reverberation, most of the time. Your options range from a very clear, unobtrusive synthetic reverb, just to add that certin feeling of "air", to samples of real halls - rich and full of character. Again, you should treat the reverb as a signal of its own (like an engineer would optimize the ambience-tracks of his recordings). Use EQs and all sonic tools you feel that are necessary (but not more [;)] ...). The crucial point is the balance between dry and wet signal. You may understand that there are hardly any guidelines for this, as it depends totally on the context.
****
I hope this gives you some starting points for your own solutions. - As this thread has turned towards a more technical POV now, I will move it to the "Mixing and PostPro"-section of our forum, where other users will most likely look for this topic first.
All the best,
/Dietz - Vienna Symphonic Library -
I think Gugliel's idea of modifying individual notes is theoretically the answer, but of course we would all go mad if we had to do that. I find it often too laborious to even instert the right crescendo articulations etc.
I have always noticed that on my recordings the VSL samples sound thinner compared to the live sections - for example my average (live) violins sections are only 5 + 5 and sound still much bigger than VSL's 14+14 or for example I recently recorded a French Horn section with just 3 horns and they sounded huge - compare that with playing chords played the with the VSL Horns Ensemble (4+4+4= theoretically 12 horns! but sound quite small).
Apart from the 3.5k dip and 12k "air" boost, that Dietz mentions too, the live recordings don't need any postproduction.
I think ulitmately we could get closer to expressive lines by having more realtime control over vibrato and dynamics (real, not crossfaded) but I wouldn't know how this could be done with samples.
-
Thanks, Dietz, for those ideas. The "raw" aspect of the VSL sounds is what makes them so adaptable and useable.
To me the most elusive quality of strings is how they will sound far darker than samples much of the time, but still not muffled as in heavy-handed EQ (the kind I usually do).
Maybe I can apply some of this expertise to a Russian Romantic I happen to have sitting right next to me now.
-
I keep meaning to try and get time in the studio to experiment further on an idea I just started to hatch ages ago.
Space Designer, or any reverb plug in for that matter, offer a functionality (hate that word) that isn't immediately obvious. By applying the reverb via a regular post-fade aux send on your mixer, real or virtual, there is a large amount of dry signal left to be thick, up front, in your face, over bright, too detailed to sit in a mix etc.
My idea, that in initial try outs sounded great, is to send the instruments to the reverb using pre-fade aux sends and then lowering the fader levels for the instruments to much lower settings. Because the aux send is pre-fade, the level of the instrument going to reverb is therefore independant to the fader position. By juggling the aux level against the dry channel fader level, you can really "sit" instruments into the mix. I have the luxury of 8 digi outs from Logic Audio and so pair off sections - Strings run through 1&2, woodwind through 3&4, brass and percussion through 5/6 and 7/8 respectively.
This system depends absolutely on you having a reverb setting that you really like the sound of. There are a couple of rooms/chambers/spaces in Space Designer that I like a lot. It also allows convolution reverbs to be appreciated to the fullest. Rather than adding some kind of a tail to a sound, the character is the reverb itself. Strings sit back and aren't so bright, flute solos come over with a real sense of space and distance. Nothing sounds overly close mic'd, which is my intention a lot of the time when recreating orchestral sounds. Sometimes, a flute is really nice in the same room as the listener, but not when playing prelude a l'apres midi.
I recommend others to have a play with this. If anyone wants to know how to set up pre-fade sends in Logic audio, then I shall be happy to furnish them with details.
Great thread, though, and totally central to any possible chance of sampled orchestras sounding convincing.
-
Well, in the thin, rarified atmosphere of this VERY quiet corner of the Forum, please let us continue this interesting discussion.
At first I thought it was a sign of hard-headed practicality as well as good ears that Dietz uses convolution with dry signal, but it is startling that it might even be preferable due to the cut off signal mentioned.
I was using the Cholakis convolutions in Vegas 6 audio with dry signal, and it seemed to work well, though I then began using gigapulse 100% wet, adjusting the so-called "perspective" because this is "pure."
However, I really LIKE adjusting dry/wet ratios! A perfect way of envisiining how close the source of sound is, compared to the ambience. So it has been like some kind of punishment not to do that because it is basically WRONG with convolution. I'm very glad to hear that Dietz ("Da Man") has not fully submitted to this punishment.
-
