Why Realism is Overrated (in the World of Sample Libraries)

When it comes to choosing an orchestral sample library, you don't necessarily need to focus on how 'real' it sounds. Jay Asher, master composer and Logic guru explains why.  

All over the internet, the emphasis in evaluating new libraries is “how real does it sound?” My response is to advise users to focus on making it sound good, not real (and no, they are not always the same).

I learned this lesson back in the early 90s when I was scoring the TV series “Zorro” (shameless promotion alert: now available on DVD). I had a smallish orchestra, about 24-piece, that I was augmenting with samples from the Emulator III and some synths, especially the Memorymoog (see Pic. 1). 

The Memorymoog was quite simply a wonderful sounding synth with a warm round sound that no samples of the day could match, I found that by doubling my real cellos with it I got a sound that I loved. Did it sound more like a real cello section after adding it? No, it did not. It sounded cello-ish. But it gave me the emotional wallop that I would have utilized a larger cello section for and to this day people still tell me how good it sounds.

MemoryMoog


As for sample libraries, the pursuit of “real” is illusory. Obviously, you do not want them to immediately sound so fake that it calls attention to that fact but very few modern libraries do. Does EastWest's Hollywood Strings sound more real than Audiobro's Los Angeles Scoring Strings, VSL's Appassionata Strings or Kirk Hunter's Concert Strings II, for instance? In my opinion, no, not in the hands of equally skilled users. But they most definitely sound different.

And this only makes sense. After all, if you listen to the same concert hall piece recorded by the Berlin Philharmonic, the New York Philharmonic, or the Boston or Chicago Symphony Orchestras, they will all sound somewhat different. Even the same orchestra may sound different with different conductors at different time periods.

If you think about recording a violin section to achieve “real”, I think you can understand why this cannot be done in a completely satisfying way. Here you are, trying to sample a bunch of really great players who bring a lifetime of experience and knowledge of their instrument: The instrument itself, which may have a very different tone, the techniques the players use which even with the same articulation cannot be 100% identical because they are human beings, and the artistry and emotions of the player.

As skilled as today's sample library developers are, there is simply no way for them to capture all this. What they can do is capture a very personal sound that different users will have a positive or negative response to. There are no rights or wrongs here, we like what we like. They also can have a workflow that again different users will like or not like, different CPU demands, etc. At the end of the today however, most users will gravitate to the libraries whose sound they like best. They may say they like it because it sounds more “real” to them, but this is certainly not empirically true and is highly subjective. What they are really saying is that it sounds better to them.

Personally, I love the sound of EW's Hollywood Strings to distraction. (In the interest of full disclosure, I work part time for EastWest as their Online Coordinator, but I will ask you to take me at my word that this does not affect what I write next.) 

EastWest's Hollywood Strings


When I first heard it, what immediately grabbed me was not how “real" it sounded, but how lush the tone was and how good the room it was recorded in sounded. I mostly get hired to compose “pretty” music and it was immediately apparent that with this library, I would be able to do that more easily and effectively for my clients. Now the very properties that draw me to it may be the same ones that draw another potential user away from it and towards another, and that is fine. There are a lot of wonderful choices and I use several string libraries, sometimes separately and sometimes blended together, depending on my perception of what the music I am composing needs. But the aesthetic pursuit of “real' is never factored in for me. They ALL sound real if I use them well and all will sound obviously fake or "synthy" if I do not. With sample libraries as with real players, you need to write for the players.

To paraphrase a line from “Jerry Maguire”, “Help THEM help YOU.” Make them sound good.



Jay is a Los Angeles-based composer, songwriter, arranger and orchestrator, conductor, keyboardist, as well as vocalist. As a composer, he is best known for scoring the New World Television series Zorro. Among the films and TV movies he has arranged, orchestrated and/or conducted are Paramount Pictures' Blame It On Rio Read More

Discussion

Thomas Goss
From the perspective of a professional concert music composer, the process of making a sound set sound good means embracing its limitations. Unfortunately, this limits the type of music one can effectively compose. A strong, beautiful idea that would be a worthy challenge for live musicians may not sound all that great on even the best of sound sets. What a shame it would be if those kinds of ideas were being abandoned in the face of technological perfection.

I think we have to accept that the search for ever-more-realistic sound sets is ultimately motivated by a need for greater expressive freedom - especially for the more developed composer who has their own internal orchestra that requires a real-time audio realization. But of course, no real orchestra will ever perform your work how you imagine it - and neither will a DAW with the most sophisticated of sound sets.

The real risk is that young composers will start composing "to the sound" - that is, treating sound sets as instruments in and of themselves, rather than merely as tools. We may be raising a generation of musicians that have little interest in the realities of dynamic balance, technical limitations, and a host of other practical concerns that every professional orchestrator must face. So I respectfully disagree with Jay that we should be making an effort to make the sounds set sound good, rather than continuing the pressure for greater realism - at least, if our goal is create a convincing and satisfying orchestral sound.
Jay Asher
You make fair points but where we disagree is that I know from experience that if you understand what real orchestral players can and cannot do and write within that, a real orchestra will indeed "perform your work how you imagine it " and perhaps even better, but not "a DAW with the most sophisticated of sound sets."

It is by definition simply not possible to recreate the sound, intellect, heart, and musicianship of 80 guys who have devoted their lives to music with one guy at a computer. So it is folly to think of it that way IMHO.

Thomas Goss
Actually, Jay, I'm a concert music composer with over 20 hours of orchestral works and arrangements performed by professional orchestras. My YouTube channel, Orchestration Online, is a resource for developing composers, and I'm the current macProVideo Sibelius trainer. And I'm about to create a 17-hour orchestration training series here on mPV.

My experience tells me that my imagination must be educated by practical knowledge of the limitations of real musicians, and enhanced by the unique possibilities within each player and team of players. So with all respect, I've heard vastly different interpretations of the same works by different pro orchestras, and indeed by the same orchestra on different days.

Perhaps I'm not making it clear. I'm in no way stating that a computer could ever truly replicate an orchestral performance (in fact, I spend almost no time whatsoever myself in such attempts, as I compose largely for real ensembles who need no such previews, and I simply don't have time). The real thrust of my counterpoint to your original statement is that there is a danger in composing to the sound, in that developing composers may become accustomed to the limitations, and start to abandon the realities. In a sense, I'm agreeing with you past the limits of your article.

So I am absent of the "folly" of which you speak. Rather, I see the unfulfillable search for perfection being driven by sound developers who are trying with all their might to make it real. They'll never get there, but I think the attempt keeps the whole point of sound sets more honest.
Jay Asher
Well, then you know well that you must write to the players as well i.e. you can not write the same things if you are commissioned by a orchestra at a small university that you can for the Boston Symphony.

I am not against the developers trying to make them sound more real.I AM against the idea the I should not i.e. write a fast passage for a trombone patch between Bb and B natural a 9th below middle C because areal trombone must go from 1st to 7th position creating a slide and against the idea that if I think doubling it with i.e. the new Solid State Symphony adds warmth but makes its sound less real that I should not do so because it makes it less real.

So we may just have to agree to disagree.

Thomas Goss
Fair enough, though the F trigger on most newer trombones will take care of that with no problem.

Sounds to me like we agree more than we disagree, actually...
Jay Asher
Indeed.
Al Johnston
First, I would like to say I am a subscriber to macProVideo. I subscribe primarily in order learn from all of Thomas Goss' tutorials, and especially his orchestration courses. But certainly I take advantage of other content, as well. I have great respect for him and his approach to teaching. Many thanks. (And I should mention that I remember the "Zorro" series way back yonder, and that is probably because of its exciting music, Mr. Asher.)

The topic raised in this article and the answers given by both the author and Mr. Goss have preoccupied me for many years. By and large, I think the discussion here is well-considered and I agree with the conclusions I read. Yet I think the issue has additional dimensions that remain unaddressed. Perhaps I am wrong, but it seems to me that the discussion so far has revolved around the needs and practices of music professionals (or budding music professionals). Or at least around those with more funds and time ... and more importantly, that have more resources other than software manuals and a knowledge base ... than I have at my disposal.

You see, unlike Mr. Goss and the author, I am not now nor have I been a professional musician - nor at this late stage do I intend to be. What's more, I don't have access to friends/colleagues that play instruments, or to tutors well-acquainted with best practices in the world of electronic music creation, nor do I feel comfortable approaching students at a nearby college. Rather, I am a writer and prefer to use Sibelius (a notation program) to write my pieces, rather than a sequencer. I am mainly interested to write chamber music and concert pieces for smaller ensembles intended for live performance, though due to the audio-visual nature of modern life wish to have a relatively realistic mock-up as well.

With the Internet, there is need for immediacy previously unknown. Recordings, not scores, are the standard media. I can't pay the dozens of musicians to play my scores, much less the recording studio it would take. Orchestras cannot afford to take on even famous "unknown" composers. If I ever hope to hear what I write, much less for people to hear what I write, then I must have a viable recording.

As most are aware though, because I am interested to write in a more traditional style I am put at disadvantage as far as playback is concerned. In general, today's programs and utilities ... even entire sample sets ... are designed and marketed to cater to music styles other than that which I am interested in. For instance, to my knowledge despite an explosion of string orchestra sample sets in recent years, including at least 6 chamber music string ensembles, there is still not one recorded on instruments using gut strings. Likewise, I cannot find a trombone sample at any price which gives me true glissando patches, much less the arcane ones I need for my trombone concerto. (Please, please tell me if I am wrong.)

So, I fall between the cracks, but there are many like me of assorted and varying degrees of talent eager to take advantage of the new technology. What to do? As any book on MIDI orchestration will attest, the first step to achieve a believable mock up is write music according to the well-established rules that ensure successful live performance. Happily also, there are many musicians on the Internet who make practical advice available to composers who present them with parts to play. And also due to the work of wonderful people on-line, scores timed to the music of the great classics depicted are available on Youtube to study. (https://www.youtube.com/user/12clar3412clar34)

But there has been no one to show me - even for money - best practices for doing convincing MIDI mock up of my music. I'm sure you understand how involved the process is; it goes far beyond whatever documentation is written down, anywhere. Youtube videos, on-line courses, and even college courses fail in this regard. In fact, like driving, it may not be possible to learn how to produce a realistic MIDI mock up (at least, of chamber music) without some private instruction by a professional. Yet there is a dearth of such paid instruction. And for notation program playback, there is none. The issue of realism ... or at least of listenability ... is moot if the industry does not make available some practical instruction for learning how best to go about achieving it.

Kontakt's manual provides a single page that tantalizingly describes the barest outline of what is required to configure a notation program for listenable MIDI mock up. The overview mentions multis, instrument banks, and MIDI CC numbers even before discussing what these mysteries are ... and then never returns to the topic again. What's worse, the Sibelius manual contains less content about the subject than that. Producers of sample libraries are well-known to be uninterested to educate their consumer. My only relief in this regard has been Peter Alexander's Visual Orchestration tutorials (Alexander Publishing), which I have paid for gladly. The in-depth discussion on assorted sample library characteristics and on reverb and recording technique for MIDI mock up is unique and extraordinary, well worth the price. However, they are not enough.

In my opinion, the danger is not so much that young composers will begin to write to the sampled instrument but to the limitations on creativity imposed by the electronic music industry itself. Not all music is "epic." Nor is it necessarily orchestral. Nor is it necessarily for profit. I derive a great deal of satisfaction from practicing my art, regardless of what others think about it. I have total artistic control. As Schoenberg is quoted to have said, “there are still a great many pieces to be written in C Major.” Those who follow what I write at my Google+ community ("Sheetmusic playback of Original Music") know that I prefer to mine paths already blazed before by others. There are so many who forge ahead, but my concern is to develop the landscape already revealed. Unfortunately, as an amateur I find I am being hemmed in rather than liberated, despite the thousands I have spent on the electronic music industry's tools, because there is a lack of concern to educate users in best practices. (Are the companies even aware that there are such?)

As my expectations diminished that sequencer programs would incorporate better notation capabilities, I have worked long and hard over the past 15 years to develop better techniques to achieve listenable MIDI mock up for notation programs. I have posted some 5 hours of free in-depth tutorials in this regard. I do not claim that the result is the ultimate in "realism." However, playback achieved using these methods go far beyond playback of any scores composed without them (whether measured in terms of sound produced, sensitivity to notated indications, or "believability").

But again, I am not a professional. It is the professional that should be making this information available to anyone using their product(s), not some user. I am not privy to inside information, nor to a team of experts. The going is slow. But it is non-existent elsewhere. I have waited for years wanting to avoid re-inventing the wheel, but I began to realize that the wheel had never been invented, nor was it about to be. All sound sample libraries released in recent years are geared to map onto a MIDI keyboard, with the intention of using a sequencer for MIDI mock up. Whatever capabilities for playback built into notation programs are not designed to compete. But I cannot accept that I must double the effort just to write and then to hear my music.
D Young
Well, I have struggled over the years with promises of realism with this fx box or that sound sample and mostly found that I could not emulate the sound that I wanted to hear. I came to the realization that if you're doing pop music then achieving the sound is simple, given that the exact sounds that are used in this genre are available to anyone and are generally performed via the same tools most have. If you're doing other genres then things get more challenging.
Our ears are certainly intimate with the sounds we have heard all of our lives. When they are imitated it is much like hearing a computer generated voice on an answering system.....something not quite right. I think that if you want a true rendition of your music you need to use that which is envisioned to be the ultimate performance medium. Now given that the orchestra is not at everyone's disposal, that will most certainly pose a significant limitation if that's your medium. Having said that, as has been alluded to, it is workable to use a blend of both samples and real players. If you want to really make it happen you need to finance for it. I will tell you that in my experience it was worth it. No one can take anything away from you when you have the power of pros backing your sound. If you can throw a few musicians into the mix, along with some samples, it can bring a useful product into fruition. To go the whole route with live players is sublime.
The samples are really quite amazing these days but it also depends on what kind of music you are doing. If you're doing action scenes with crash/boom, unrelated chord changes every two beats and lots of piccolo and string runs, most can be fooled into thinking it's a real orchestra regardless of how boring that kind of over done orchestrating is now. Note that the latter is a reference to having to write to the scratch track to the point that there seems to be little creativity left to the composer who would like to try something different to achieve the effect.
I have also noted that many of the regarded sample libraries often hide the very instrument you want to hear behind a barrage of other instruments to create "realism". It's like the Zebra effect when lions are hunting... for the listener. If you listen very carefully you can hear the unreality of the sound but given that the track is going to be used in a multi-media environment with dialogue, spec effects and visuals.....who's going to notice? This brings about the point of intended use.
If one's music is for pure listening then the acoustics have to be exceptional as there are no distractions here and the sound quality is of prime importance. For this I have some close associates who I call when I really need to deliver on an emotionally impactful performance. For those that haven't engaged players, I suggest you call your musicians guild and contact people in your local pro orchestra and see if you can strike up an agreement. It doesn't have to be wallet stripping and will provide you with a lovely experience that you will most certainly enjoy. They don't care if they are recording in your living room. I can't stress this enough. If you want to produce pro you have to hang with pros. What they will give your ears in a one hr. session will blow you away or at least inform you. They can also be of great help in providing perspectives. If it cost you as much a $100 bucks for the hr. it's well worth it. I've had players come in for $50 and in these times many are happy to do it. Even a single part will make a big difference. I'm a French Horn player among other things and nothing can compete with my sound except another player.....even just long tones can't be beat. When I mix in other horn samples it becomes epic though I have also done the same multi-tracking my own playing. There is just a life and a tone there that defies sampling.
When you put out a really great sample sound and then hear the real thing beside it.....there is no choice....real does it. Bring in even one player to help out on a lead part and you'll be smiling. You'll also have a new friend....and if they love your music you'll get something else on another level....and you can't buy that.

Want to join the discussion?

Create an account or login to get started!