Friday 1 April 2016


What is wrong with synthetic instrument performances?





Before getting to the core of this blog I would like to relate a short story.  Some three years ago my wife bought me a present, a manuscript copy of "Lambert's Clavichord" a series of short keyboard pieces by Herbert Howells. In turn I made her a gift of a recording of the music on harpsichord as she prefers the sound of this instrument to the piano, also it is closer to the sound desired by the composer. The harpsichord was a sound sampled instrument from Kontakt and one which I find more than satisfactory for the purpose.

I put the recording on YouTube and it has amassed over a thousand visitors which is a pleasant surprise. A few days ago I had an e-mail which said



"Is this some sort of MIDI thing"?                           



I wasn't certain if this was a criticism but it certainly had the edge of an unfavourable response. I felt uneasy with this as I had spent many months planning the performance, spent considerable time working through the pieces, played the music in real time and then corrected the errors by a mixture of rerecording and some note editing via MIDI. There was probably no more editing on this than happens with most studio recordings. These days synthetic harpsichords are used in concert halls as their tuning remains true, and various turnings are possible with a touch of a switch, so what was the problem for this listener?



The use of synthetic instruments raises many questions and responses and I have experienced a number of these first hand. This is hardly surprising as I am particularly interested in the use of this medium.



Most musicians will be familiar with Sibelius score writer, if you use the basic MIDI set and select violin you can have a short period of entertainment or experience sheer hell as you experience a strange sound unrelated to any instrument known by man.  Its limitations are huge. I am not going to compare this string sound with others, strings are notoriously difficult to emulate well. What I will say is that if you have access to samples of glissando, pizzicato, sfz attack etc, and use these with care the result is more truthful and easier on the ear. It comes down to this, the more time and thought you put into the final product the better the result. Of course a satisfactory result means considerably more work on the project, and why would one do this if there is a competent violinist waiting to perform your work around the corner? I don't think I need to expand on that comment if, like me, you write works which explore some of the more contemporary aspects of music.



The problem of synthetic sounds was discussed between Giorgio Sollazzi and myself some weeks ago as I tried to make an arrangement of his Kantata, his reply was pertinent, I hope my translation does justice to his answer:



When I use sampled sounds I always try to give them an electronic soul, I never want to simulate an orchestra and instead I want to feel how dramatic or ironic or sarcastic synthetic music can be. You know the difference if you pay attention to the attack, in MIDI it is always the same....the envelope is always different.



This gets to the heart of why some synthetic sounds simulate better than others. So if Kontakt release a version of the clavichord I hope they will put finger pressure samples in their collection before I rework “Lambert's Clavichord”.



Another response I had to a Kontakt score (this time a string quartet) was more positive but with this proviso, “you don't get the vibrato you need with sampled instruments” (Anthony Littlewood). In this case the problem was different in that I wanted a sound that was static or still as possible, sans vibrato. The sampled sounds of course will have some vibrato, and if desired it is quite an easy task to add vibrato to a sample with sine waves. So the problem of getting what you want cuts both ways.



The response by Sig. Sollazzi raises a big issue, that is that the mind seems to prefer the “messiness” of certain types of information. Perfect environments are sterile and unwelcoming. This leads to the question of perfect performances on disc which is too large a concern to deal with here, but on a personal level I prefer a radio broadcast of a live concert, and e.g. a performance of “The Goldberg Variations” with occasional slips is no great hindrance to the enjoyment of the work as a whole.



I always think that artists have a more liberal approach to technology than musicians, while the photography v paint argument still rumbles on, it started in the late1830’s, the generally held belief is that technology and art are two sides of the same coin.



There is a Zen approach to examining is dividing the issue of synthetic sounds into the technical and aesthetic considerations, here is an incomplete table of my thoughts. It will constantly modify as enthusiasts and professionals work side by side to make more demands on the possibilities in synthetic sounds.






4 comments:

  1. This today especially needs to be shared with original sheetmusic playback here on G+. They are all about the mix, and composition,it is very well run,introduce yourself to Al Johnston Ken. I think you have summed up much more elegantly than I would at the so called accuracy of current cinematic realizations, midi's passed off as authentic,etc.
    On the finer points, control of vibrato and actually how many strings you are using in your mixer panel seems to be the real difference to the trained ear. Now if someone has never met a stringed instrument in person, one may not know the difference. I loved all the old Moogs,Yamaha sting orchestras,wave generators of late 70's I played around with.That and oodles of LP's of course for the big 3 of that,Williams,Varese and Tomita.
    Played thru on some through electronic manipulation of my viola& guitar as well. However, not once have I fooled myself into thinking it was remotely organic. I think Giorgio makes the most valid point on this I have heard recently perhaps treat it as a separate audio from the scores end purpose &you have notate the audio as a realization.Some families especially string,voice and depending on range winds, and have heard nothing remotely accurate in brass section to humans. Live is best depending on audience whether recorded or radio, happy accidents be damned, sometimes the mistake is more interesting. I am a loyal fan of Avid, but the mockup they have on Finale for a cross upgrading sound library is so much better. I seldom worry about the end product being that much accurate as it's own end unless it is something to practice with for someone I teach,they are beginning guitarists though,so strum patterns are just a more interesting metronome. I can't play ,so now I teach ;)
    If you can program the sensitivity that'd be great for keyboard. However,seldom does string music transfer over to something largely percussive in nature, unless your talking organs, and lacks sublety.
    I give you Busoni's sledge hammer treatment of the immaculate Bach Chaconne for example for piano. Even in Serkin's hands it was regrettable, however,early keyboard music including the infantile stage of pianoforte in early years and before sounds quite nice on guitar due to a better match of controlled sustain.

    ReplyDelete
  2. Once I was chided by a friend for calling a MIDI performance of one of my pieces "performance". Her position was that what she heard was not performance. Fortunately, she cpuld not come up with a definition of what she heard was. If you consider music to be a deliberate act of organisation of sounds by one more human beings, the act of producing these sounds by any means becomes performance. The logical definition of ''music'' leads to the fact that the realisation of a piece using a set of computer generated sounds is a "performance" of the piece - whether it was written for a computer or not. There is no difference between MIDI performance of a piece written for a string quartet and the performance of the piece as a piano reduction. Just another point of view,

    ReplyDelete
  3. Thank you both for your views, I like your term "organic" Jeff. To widen Nurtan's point it would be insane to suggest that a recording is not a performance because it has been processed. I have been thinking about our senses in relation to this question, much of the information in a live performance is visual, timing and articulation is often demonstrated by the performer. More on this idea later.

    ReplyDelete
  4. A couple of months ago I wrote either to this site or to the general G+ site moaning about controlling vibrato on sampled sounds from string instruments, and I am still moaning about it. I was particularly struck by the problem when I listened to the rendering of Nurtan's piece for violin and piano. Go to bar 120 where there is an ff low G on the violin, and you will hear a lovely vibrato which simply shouldnt be there! OK, that is only a problem with one note, but there are several bigger problems. Here's three which pertain to my own setup Cubase 7a with Halion sounds, but you hear the faults in other pieces all over the net:

    1. Very often you want to have notes senza vibrato. You are not given the option of turning it off.

    2. Violinists usually start a long note without vibrato and only start it after a definite delay. This facility, which was commonplace a decade ago, is now absent.

    3. In an ensemble of string instruments, they all have vibrato, but the vibrato is the same for all the instruments, so that in effect, a chord has vibrato. This invariably damages the music.

    In the old days ( that is fifteen years ago) when we had to synthesise our own sounds, the available software could handle all the difficulties perfectly well, and sometimes I wish I had it back. But I suspect that I'm sure others have encountered my moans, and if they can help by advising me of different equipment that I could use, I'd be very grateful.

    ReplyDelete