First, I do know that there is nothing to beat actually 'playing in' parts into a score for that 'human' playback feel. However, my work flow remains the MIDI route and I want to achieve the very best performance possible through this route.
For those who 'play-in' parts into their notation editor/sequencer there is a tool to tighten the performance timing, the 'quantization' tool. It wouldn't really be an option to go through even a 'modestly sized' score and quantise each individual note one by one. We need simply to select the whole score, however large and involved, select the 'Quantization' tool and the tolerance value and the tool does the rest .. voilà .. a well-timed score with random timing 'imperfections' and as such retains the 'Human Performance' factor. Well, what about the exact opposite; the other way around, i.e. Take a MIDI score with it's unnatural timing perfection and de-time it randomly to humanise the performance.
I have read several threads and articles, including our own DP Dan's excellent tutorial. However, I still have a myriad of seemingly conflicting brain-waves and unanswered questions.
From what I have read, I believe there are two ways to achieve the same effect. The first is to physically move notes a few ticks out of time. The second is to leave notes as they are but to alter the attack on notes randomly.
(1). Are the two methods I describe above really the same thing or are they indeed two different methods?
(2). If they are different methods which is considered the most effective?
(3). There are tools in Finale (Randomize) and Sonar (Humanize and Random Time). I have tried these tools and when I examined the scores there was no change; the notes were exactly as they were. I have been told that the tools randomises the attack and not the timing as such. When listening I could hear no difference. Should I either be able to see or physically hear any difference?
(4). Are there other industry standard or professional tools that will either randomise 'de-timing' of notes and/or attack to 'humanise' a MIDI score i.e. Select the whole score, select the 'De-time Tool' and voilà, every note is 'de-timed' randomly to create a humanised performance?
(5). Does anyone actually visit every note in their score to randomly de-time or just a cross section of notes in each part?
(6). Sonar uses 'ticks' (960 = quarter note) as its smallest timing unit of measurement and Finale uses EDUs (1024 = quarter note). Too much adjustment would put the work out of time whilst too little would make it still sound too perfect. Is there such a thing in the world of the professional audio engineer of what is considered the optimum de-time?
(7). If it's just a cross section of notes de-timed, roughly what kind of percentage does the trick, less than half, less than quarter, just a few notes here and there?
(8). In GPO, I understand that the instrument sections such as Violins 1, Violins 2 etc. have been prepared with random attack which is fantastic. If using just one section with GPO, say violins 1, am I right in saying that we should have a nicely 'humanised' performance and that no further de-timing should be necessary.
(9). If (7). above is incorrect, why would it need further work?
(10). When using more than one instrument section with GPO, say a string orchestra, as each section has been 'prepared', is it still necessary to de-time each section?
(11). I have read that some people add a solo violin (or other instrument?) to a score to give the strings more presence. I have experimented with this and, personally, for me, I dislike the effect and would prefer not to do this. Do you add a solo instrument every time irrespective?
(12). If you do add a solo instrument irrespective, doesn't processing the score through de-timing the attack or actual timing of the notes alone create a sufficiently 'warm' 'humanised' performance without adding other instruments?
Well, that's it for now. If anyone can help with any of the above questions I would be really grateful.