--- In logic-users@yahoogroups.com, Steven Rowat <steven_rowat@...> wrote: > Fast forward to trying to do the same thing with headphones on, > hearing a drum-beat (from the computer), and my own instrument > coming back to me 20 milliseconds after put my finger down. What do > I sync my internal model to? From my point of view, 20ms is mostly unplayable, and I'd be reducing the buffer size, or hardware direct monitoring, and encouraging the performer to learn to play to the latency. If that proves impossible, I'd provide a fully analog monitor path. > But as a musician recording myself [...] > So, when I do play it back, I want to know what setting of delay > will give the best chance of success There is no good setting if you're expecting the compensation to adjust for your (variable) built-in compensation. You simply CANNOT know how good your built-in compensation was, and when you were or were not doing it. You might as well cut the result up and nudge it around until it sounds right, but that can be very tedious, and is a slippery slope. That's why I go for compensation that simply reproduces what was monitored. It's reliable. > You said: > >B - the musician should be strongly encouraged to play to the > >monitor signal. > This makes it easier for the enginner/producer, but I suspect it > does not make for a good performance. If the latency is low enough, any reasonably competent musician can compensate pretty well and play to the monitor signal. What constitutes "low enough" depends on the musician, but we're probably talking around 10ms or under. Remember, as has been pointed out, musicians are perfectly capable of playing to a 10ms latency when they are 10 feet from their amp. Sometimes, monitoring through phones, just being able to *see* the amp at the right distance can help the brain out, making the monitor latency feel natural. > for myself, so I'm going to start some tests playing along with a > click-track using MIDI/EXS software monitoring and a large buffer OK, so far we have been taking about recording audio. Now we're into MIDI, and aside from MIDI input latency and audio output latency, you have the additional problem of MIDI jitter, which means the latency is also variable from note-to-note. In my experience, this is FAR harder to play consistently to. And when using software instruments, there is also another issue relating to the way Logic force-compensates for the output buffer for audio instrument tracks, which means the audio instrument will play back earlier than you monitored it while realtime-recording (by the output buffer amount). This is the basis of the "sample accurate" claim for audio instruments. You may like this. I don't - it messes with my timing after the fact. Once again, what I heard is not what plays back. This aspect deserves yet another discussion and a page on my website. > But it will really simplify things if I can figure out which one - > source or monitor - I sync to most closely while playing. I suspect you'll find that it's all over the place, depending on tempo, time of day, tiredness, coffee, instrument, whatever. Don't rely on what you measured yesterday... I went down this route a few years back and that's what I found. My solution was to take the producer's position on what is "accurate" (the monitor signal), and put my energy into reducing the monitor latency, analog monitoring where possible, and where not, learning to play in time with the monitor signal. At a reasonable latency, it's really no harder than a bass player a few feet from their amp learning to sync with a drummer. John Pitcairn ---------------------------------------------------------------------------- - LC Xmu Logic/Mackie Control emulation & management, LC Xview software LC/MC display, Logic environments & stuff... Opus Locus - http://www.opuslocus.com ---------------------------------------------------------------------------- -