Is something wrong with dynamic cross-fades? Like 12dB of gain change.

[BruttoBello]

Hello all,

Most of my orchestra libraries which have dynamic cross-fades seem like they are lacking the true dynamic range of the sections. For instance, several libraries (East West, Sonic Implants, Dan Dean) only seem to change by around 12 dB as they cross fade. The timbre changes from piano to forte, but the volume is limited.

What is the normal range for the change from ppp to fff? Am I using the libraries wrong? It seems that the velocity is bypassed, so the mod-wheel should be controlling the entire range from ppp to fff, which should be more than 12dB if it is meant to recreate the correct change in real life of the orchestra sections. This seems to happen across multiple libraries and with all sections.

Any help?
Thanks, Nick

Click to read more...

 

[chibear]

In short, yes I have dealt with that for quite awhile. The way I get around is to use the cross fades (if I can, programmed to CC#11) to manipulate timbre and then make up the volume using CC#7. Works fine for me.

 

[BruttoBello]

Yeah, that makes sense, but real players don’t have a volume knob. They play harder. That tonal shift and volume is recreated by sampling multiple different volumes from the sections and is then reproduced using samples which are recorded with players playing harder or softer as appropriate. Shouldn’t a single knob controlling the fade between those dynamic levels be enough to change timbre and volume just like the real thing?

 

[Kent]

This is such a complex issue, but two basic things to consider:

1. Compression. Pretty much any recording you hear is going to be artificially limited in DR. Why not make things a little easier for the ‘performer’ from a VI standpoint?

2. Proximity effect. The closer the mic is to the sound source, the more DR it will pick up; conversely, the further away it is, the more ‘even’ it will be thanks to the inverse square power of propagating sound waves and the simultaneous psychoacoustic and actual reinforcement of reflections.

 

[José Herring]

Hello all,

Most of my orchestra libraries which have dynamic cross-fades seem like they are lacking the true dynamic range of the sections. For instance, several libraries (East West, Sonic Implants, Dan Dean) only seem to change by around 12 dB as they cross fade. The timbre changes from piano to forte, but the volume is limited.

What is the normal range for the change from ppp to fff? Am I using the libraries wrong? It seems that the velocity is bypassed, so the mod-wheel should be controlling the entire range from ppp to fff, which should be more than 12dB if it is meant to recreate the correct change in real life of the orchestra sections. This seems to happen across multiple libraries and with all sections.

Any help?
Thanks, Nick

These are very old libraries and the method of sampling was much different which lead to not very much dynamic range in the patches. As has been mentioned the way around it was to use cc11 plus cc1 carefully to mimic the dynamic range.

The next best thing to do is to get a more modern library.

 

[BruttoBello]

Thanks for the input. I've tried more modern libraries. For instance, the Spitfire stuff uses both CC11 and CC1 to control volume (CC11) and dynamic level cross-fades (CC1). Problem is, that with that setup, I could (in theory) play a pp sample (CC1 set at value 5, CC11 set to 127) apparently louder than a ff sample (if played with CC1 at 127, CC1 at 5). Because CC11 controls the overall volume from 0dB to full volume, and CC1 only changes the volume by about 12dB, all that is needed is to adjust CC11 to compensate (or overtake) the 12dB change.

Most libraries I've tried seem to let CC1 affect the volume somewhat, but not completely. Why is there need for a second control (I mean aside from CC7 - volume)?

 

[Pianolando]

I think what you are experiencing is that the sample library developers have raised the volume on the softer samples compared to the louder ones to make the “instrument” more uniform and more easily playable. The dynamic range of a violin section, or a single violin, is vastly much more than 12 dB. It’s not an issue of it beeing older libraries per se, as Hollywood strings has more dynamic layers than most new libraries, I think it is 5 for legato. It’s a scripting issue.

I too wish that the softer samples were actually softer but you can ride CC11 to compensate.

 

[BruttoBello]

Thanks Pianolando. I guess it makes sense that the instruments are more easy to play. It feels like there is a compressor on the channel.

With either the low dynamic range on CC1 or the dual CC way of doing things it just seems weird that neither represent what happens in real life when a player/ section swells from the quietest to the loudest. To me, that would be more playable.

Could it be that people don’t find 127 values enough?

Sorry to belabor the point, but it doesn’t make logical sense to me. I am about to re-program my libraries to feel better to me, but wanted to input from the forum first.

 

[chibear]

Yeah, that makes sense, but real players don’t have a volume knob. They play harder. That tonal shift and volume is recreated by sampling multiple different volumes from the sections and is then reproduced using samples which are recorded with players playing harder or softer as appropriate. Shouldn’t a single knob controlling the fade between those dynamic levels be enough to change timbre and volume just like the real thing?

Ya, as a real player who played horn professionally for almost 50 years and taught the instrument at the university level for over 20 years I disagree with most of that statement.

Fine educated professional players are able to deal with timbre and volume separately. Thus, speaking in MIDI terms, a player will be able to play colours from the middle dynamic layers at high volume levels, the so-called “noble forte”. Listen to a German recording of the horn solo near the beginning of the finale of Brahms 1st Symphony as an example.

Conversely an example of using colours from higher dynamic layers played at a low volume would be a French recording of Ravel’s Pavane for a Dead Princess. This technique carried to the extreme is termed “cuivre” which in the early part of the 20 th century was a technique that called for a change in embouchure shape to produce a brighter sound.

In reality the introduction of “edge”, “brightness”, etc into a brass sound actually decreases the projected resonance so in essence what is perceived as louder is actually softer when sound pressure is measured at a distance of 20 feet. Arnold Jacobs, Jay Freedman, Vincent Chicowitz(sp?), and Adolf Herseth did studies of this phenomenon in the late 60’s at Roosevelt and Northwestern Universities. As a student I was privileged to participate in some of these studies. Documentation probably exists at both institutions.

What you are getting in most sample libraries is an engineer’s and producer’s conception of how a particular instrument is played. Often this has little to do with how a player actually approaches the instrument. Thus you have velocity levels (timbre) inextricably tied to volume levels, with some of the higher levels going beyond what any sane player would conceive let alone perform. In fact, of the libraries I have used, the ONLY library I find the top velocity/cross fade/whatever samples usable is the Chris Hein libraries. The upper levels of the rest I would never consider playing and give my horn students hell for sounding like that.

In the past I have been quite vocal about this and like to think that what you are experiencing is actually evidence some developers are actually taking my comments to heart.

The fact that people are still insisting on locking volume to timbre IMO shows a weakness in MIDI pedagogy which needs to be corrected in order to encourage developers to continue in a more logical and artistic direction. Please don’t take this as a personal attack because I am sure many more people will agree with your problem than my solution.

 

[Bernard Duc]

Ya, as a real player who played horn professionally for almost 50 years and taught the instrument at the university level for over 20 years I disagree with most of that statement.

Fine educated professional players are able to deal with timbre and volume separately. Thus, speaking in MIDI terms, a player will be able to play colours from the middle dynamic layers at high volume levels, the so-called “noble forte”. Listen to a German recording of the horn solo near the beginning of the finale of Brahms 1st Symphony as an example.

Conversely an example of using colours from higher dynamic layers would be a French recording of Ravel’s Pavane for a Dead Princess. This technique carried to the extreme is termed “cuivre” which in the early part of the 20 th century was a technique that called for a change in embouchure shape to produce a brighter sound.

In reality the introduction of “edge”, “brightness”, etc into a brass sound actually decreases the projected resonance so in essence what is perceived as louder is actually softer when sound pressure is measured at a distance of 20 feet. Arnold Jacobs, Jay Freedman, Vincent Chicowitz(sp?), and Adolf Herseth did studies of this phenomenon in the late 60’s at Roosevelt and Northwestern Universities. As a student I was privileged to participate in some of these studies. Documentation probably exists at both institutions.

What you are getting in most sample libraries is an engineer’s and producer’s conception of how a particular instrument is played. Often this has little to do with how a player actually approaches the instrument. Thus you have velocity levels (timbre) inextricably tied to volume levels, with some of the higher levels going beyond what any sane player would conceive let alone perform. In fact, of the libraries I have used, the ONLY library I find the top velocity/cross fade/whatever samples usable is the Chris Hein libraries. The upper levels of the rest I would never consider playing and give my horn students hell for sounding like that.

In the past I have been quite vocal about this and like to think that what you are experiencing is actually evidence some developers are actually taking my comments to heart.

The fact that people are still insisting on locking volume to timbre IMO shows a weakness in MIDI pedagogy which needs to be corrected in order to encourage developers to continue in a more logical and artistic direction. Please don’t take this as a personal attack because I am sure many more people will agree with your problem than my solution.

The separation between timbre and volume is only true to some extend. Yes, there is space for some timbral adaptation but despite that timbre and volume are still strongly connected. A string section playing piano will never sound like a string section playing forte and even the french horn playing cuivré can only happen at some dynamic levels as you know well, and in the finale of Brahms 1st the horns don't get that loud (which is why Brahms uses such a light accompaniment, to allow the horns to keep a very round sound and make them sound louder in the context). The reason why CC11 exist is to give the mockup producer this kind of flexibility when they need it, and to avoid going through noticeable crossfading when starting and ending notes, but it's a very good thing that volume and timbre are linked, as that's how the instrument would more or less play without any special indication.

As for the horns playing FFFF in sample libraries, it's not a technique really used in classical music (but I would argue that Chicago under Solti could sometimes sound like that), but it's often used in film scores. If you record in LA or London the players can absolutely play like that and brass players in general tend to be fantastic at extreme dynamics at the extreme of the range. But it is extreme playing that few players master, and that's why libraries recorded elsewhere don't get it right most of the time.

And for the OP: some libraries might use a bit of compression to help with crossfading, but it's also simply a matter of the libraries not recording enough dynamics. 3 dynamics is by far not enough for most instruments, and even 5 is still quite low for many instruments. The virtual instruments of the future will either have way more dynamics or will use modelling to smooth out and extend the dynamic range. And the limit of 127 steps is absolutely not a problem to have a big enough dynamic range.

 

[BruttoBello]

Aha, I found a video () showing some Brass crossfades. Below is a screen shot of the audio. I think it just varies by library. Check out how some of the libraries hardly change in volume (Berlin Brass specifically), while others (Spitfire, Cinematic Studio, Junkie XL) exhibit a bit more of what I would expect when moving from pp to ff.

 

[Bernard Duc]

Aha, I found a video () showing some Brass crossfades. Below is a screen shot of the audio. I think it just varies by library. Check out how some of the libraries hardly change in volume (Berlin Brass specifically), while others (Spitfire, Cinematic Studio, Junkie XL) exhibit a bit more of what I would expect when moving from pp to ff.

It's a great example. Listening to that it's obvious that some libraries do sample "compression" (they don't actually use a compressor, just turn up the volume of the quieter samples) while some others keep the original dynamic range. But that's why CC11 exist, this way you can all make them behave the same way (as long as the dynamic range has actually been recorded).

 

[chibear]

The separation between timbre and volume is only true to some extend. Yes, there is space for some timbral adaptation but despite that timbre and volume are still strongly connected. A string section playing piano will never sound like a string section playing forte and even the french horn playing cuivré can only happen at some dynamic levels as you know well, and in the finale of Brahms 1st the horns don't get that loud (which is why Brahms uses such a light accompaniment, to allow the horns to keep a very round sound and make them sound louder in the context). The reason why CC11 exist is to give the mockup producer this kind of flexibility when they need it, and to avoid going through noticeable crossfading when starting and ending notes, but it's a very good thing that volume and timbre are linked, as that's how the instrument would more or less play without any special indication.

As for the horns playing FFFF in sample libraries, it's not a technique really used in classical music (but I would argue that Chicago under Solti could sometimes sound like that), but it's often used in film scores. If you record in LA or London the players can absolutely play like that and brass players in general tend to be fantastic at extreme dynamics at the extreme of the range. But it is extreme playing that few players master, and that's why libraries recorded elsewhere don't get it right most of the time………

A couple more misconceptions there that should be addressed. I’ll let a string player comment on manipulation of colour on their instrument. However:

As originally conceived, cuivre can apply to any dynamic level. It is a colouristic instruction. Generations of improper use of the term by unknowledgeable composers has reduced it to what you imply.

The thinness of orchestration of the Brahms is irrelevant to the volume played. While the solo is only marked forte, stylistically it should be played as large as possible without inducing edge. Volume is limited by player, instrument and the register.

Your comment concerning the brass playing in Chicago under Solti has probably caused an earthquake in Chicago as all those players roll over in their graves. They were the people from whom I learned the conceptions and techniques to accomplish them. They took great pride in being able to control colour over the entire dynamic range. The London recordings are poor representations of the orchestra at the time to the extent that principal trumpet Adolf Herseth refused to allow the recording of Mahler 5 into his house because the sound quality disgusted him so. As a student I also played extra in the orchestra many times during Solti period so can attest to that amazing sound that was so mishandled by London engineers and producers.

Uniquely LA and LONDON studio players can produce the sound you refer to often, and without injury (that I know of). Don’t confuse that with the idea that they like it or do it naturally or voluntarily. I think Jay posted a video concerning instrument colours during another of these discussions where the horn player very diplomatically commented on this.

 

[Bernard Duc]

A couple more misconceptions there that should be addressed. I’ll let a string player comment on manipulation of colour on their instrument. However:

As originally conceived, cuivre can apply to any dynamic level. It is a colouristic instruction. Generations of improper use of the term by unknowledgeable composers has reduced it to what you imply.

The thinness of orchestration of the Brahms is irrelevant to the volume played. While the solo is only marked forte, stylistically it should be played as large as possible without inducing edge. Volume is limited by player, instrument and the register.

Your comment concerning the brass playing in Chicago under Solti has probably caused an earthquake in Chicago as all those players roll over in their graves. They were the people from whom I learned the conceptions and techniques to accomplish them. They took great pride in being able to control colour over the entire dynamic range. The London recordings are poor representations of the orchestra at the time to the extent that principal trumpet Adolf Herseth refused to allow the recording of Mahler 5 into his house because the sound quality disgusted him so. As a student I also played extra in the orchestra many times during Solti period so can attest to that amazing sound that was so mishandled by London engineers and producers.

Uniquely LA and LONDON studio players can produce the sound you refer to often, and without injury (that I know of). Don’t confuse that with the idea that they like it or do it naturally or voluntarily. I think Jay posted a video concerning instrument colours during another of these discussions where the horn player very diplomatically commented on this.

I didn't say that the Solti brass were bad... simply the CSO under his direction often played their FF with way more strength and edge than any other orchestra at the time, and I'm talking about all their recordings, no the London ones in particular. If you listen to the last movement of Bruckner 8th with Solti leading the VPO and the CSO you can hear the huge difference in sound . The CSO had a much more focused as well as sharper sound. The other big difference is that the CSO would ease much less on long notes while the other orchestras would taper more those notes.

And there are limits to the timbre: in this video for example Jim Thatcher (who played on plenty of LA sessions) does say that cuivré has to be at least forte. He says that if he sees cuivre but it's mezzo piano he will actually play it stopped because he can't play mezzo piano and cuivré (and he can play!) And he also says that if it gets louder at some point it gets cuivré anyways. Of course he can try to control it or encourage it.

And yes, the FFF horn playing is as I said an LA and London thing, and while I don't know the players in LA (I've been to a couple of big sessions but just as a visitor), I know that here in London the players don't mind playing that. They can do it and take pride in having "the" sound, they also use the larger bore horns for that as you know. And this explain why sample libraries try to go after this sound, it's a very powerful sound that can be heard in countless movies.

 

[BruttoBello]

Ya, as a real player who played horn professionally for almost 50 years and taught the instrument at the university level for over 20 years I disagree with most of that statement.

Fine educated professional players are able to deal with timbre and volume separately. Thus, speaking in MIDI terms, a player will be able to play colours from the middle dynamic layers at high volume levels, the so-called “noble forte”. Listen to a German recording of the horn solo near the beginning of the finale of Brahms 1st Symphony as an example.

Conversely an example of using colours from higher dynamic layers played at a low volume would be a French recording of Ravel’s Pavane for a Dead Princess. This technique carried to the extreme is termed “cuivre” which in the early part of the 20 th century was a technique that called for a change in embouchure shape to produce a brighter sound.

In reality the introduction of “edge”, “brightness”, etc into a brass sound actually decreases the projected resonance so in essence what is perceived as louder is actually softer when sound pressure is measured at a distance of 20 feet. Arnold Jacobs, Jay Freedman, Vincent Chicowitz(sp?), and Adolf Herseth did studies of this phenomenon in the late 60’s at Roosevelt and Northwestern Universities. As a student I was privileged to participate in some of these studies. Documentation probably exists at both institutions.

What you are getting in most sample libraries is an engineer’s and producer’s conception of how a particular instrument is played. Often this has little to do with how a player actually approaches the instrument. Thus you have velocity levels (timbre) inextricably tied to volume levels, with some of the higher levels going beyond what any sane player would conceive let alone perform. In fact, of the libraries I have used, the ONLY library I find the top velocity/cross fade/whatever samples usable is the Chris Hein libraries. The upper levels of the rest I would never consider playing and give my horn students hell for sounding like that.

In the past I have been quite vocal about this and like to think that what you are experiencing is actually evidence some developers are actually taking my comments to heart.

The fact that people are still insisting on locking volume to timbre IMO shows a weakness in MIDI pedagogy which needs to be corrected in order to encourage developers to continue in a more logical and artistic direction. Please don’t take this as a personal attack because I am sure many more people will agree with your problem than my solution.

I get this, and it probably makes sense to even have an additional CC to control the timbre for specialty tones or performance techniques. To me, it seems like the dynamics cross fade control should control dynamics though, right? And in thory those dynamics should be replicated at their relative volume to one another, rather than somewhat “compressed” where quiet parts are played back louder.

 

[chibear]

I get this, and it probably makes sense to even have an additional CC to control the timbre for specialty tones or performance techniques. To me, it seems like the dynamics cross fade control should control dynamics though, right? And in thory those dynamics should be replicated at their relative volume to one another, rather than somewhat “compressed” where quiet parts are played back louder.

One of the old libraries, I am told, had an “edge” layer that could be faded in and out on a separate track and accomplish that part. That was several years ago and when I looked around I couldn’t find it.

What I have tried to illustrate and strongly object to is the assignation of a certain colour to a specific volume level. If I were king of MIDI I would control the timbre and cross fades with one control and manipulate volume with another (kind of what I end up doing now for the most part). I realize in less-than-educated hands this would produce some amusing results, but for me it would exactly fit my work flow. My Chris Hein libraries work very close to this.

Your way of cross faded timbres throughout the dynamic range while very convenient would also dictate to a performer what colour they must use which is in fact their job. And worst of all, students will hear the changes as programmed and imitate them without any logical forethought.

I do, however, agree with your point concerning the compression of the lower dynamic layers as most people will accept that level as soft and the dynamic range of the composition will be further compressed.

 

[Duncan Krummel]

The absolutely brilliant violinist, Peter Sheppard Skaerved, with whom I’ve had the pleasure to work with a couple of times to premier new works, once demonstrated for me the playing of sul pont high up on the fingerboard, and the playing of sul tasto close to the bridge. It was an incredible display of bow control.

Sample libraries, as has been stated, are an impression of what one person expects. I think there is an inherent problem there that’s often misunderstood or ignored, and simply impossible to get around. At the same time, there’s a beauty to it. If you have to write to the library, you have to adopt an approach outside of yourself.

I’m not sure whether this contributes to the thread or not, but I’ve been enjoying @chibear’s explanations and was reminded.

 

[José Herring]

Thanks Pianolando. I guess it makes sense that the instruments are more easy to play. It feels like there is a compressor on the channel.

With either the low dynamic range on CC1 or the dual CC way of doing things it just seems weird that neither represent what happens in real life when a player/ section swells from the quietest to the loudest. To me, that would be more playable.

Could it be that people don’t find 127 values enough?

Sorry to belabor the point, but it doesn’t make logical sense to me. I am about to re-program my libraries to feel better to me, but wanted to input from the forum first.

Not as simple as that. To a certain extent you have to use your ears and kind of get a feel for how to use the faders with each library. There's only so much crossfading can do. Another way to look at is you use CC1 for the overall dynamic and CC11 for the subtle dynamic shades in between. Sometimes you just have to use CC11 just to make up for the fact that the developer didn't really sample that may velocity layers ect. It's a matter of feel and taste.