Improving Recording Timing Nuances by Benjamin Robert Tubb ================================== This chart converts Tempos (BPMs) from 40-330 into Millisecond delays for 1 clock tick PPQN settings from 96-1024. It is useful for maximizing your sequencers resolution PPQN settings for obtaining the best "nuance" recording within MIDI's bandwith limitation limit, i.e. one MIDI byte takes a minimum of (10 bits divided by 31,250 bits per second rate) .32 milliseconds to be sent. It can be used by the more versatile computer sequencers but also for hardware fixed PPQN sequencers such as the XPs. Because the XP-50 and XP-80 internal sequencer resolutions are fixed at 96 PPQN, a way to get around the recording bandwith for better live "timing nuance" recording, is to "manipulate" the tempo to "virtually" improve the MIDI data's timing accurancy. In other words, to obtain notably improved sequencer recording functionality at slow tempos especially, use a multiple of the "actual" tempo to record with thereby perceptably minimizing the delay times of note events. For example, if recording a piece whose actual tempo is 60 BPM, to obtain better recording PPQN resolution, use a "virtual" tempo of perhaps 240 to minimize the delay of 1 PPQN from 10.4178ms to 2.604ms. I recommend always trying to, since the ear can perceive timing fluctuations as small as from 5-10ms, use the chart to "compensate" for better PPQN resolutions based on tempo. Time for PPQNs with 1 PPQN Delay Time in Milleseconds (ms) Tempo Beat/Sec 96 120 192 240 384 480 768 1024 40 1.500000 15.625 12.500 7.813 6.250 3.906 3.125 1.953 1.465 50 1.200000 12.500 10.000 6.250 5.000 3.125 2.500 1.563 1.172 60 1.000000 10.417 8.333 5.208 4.167 2.604 2.083 1.302 0.977 70 0.857143 8.929 7.143 4.464 3.571 2.232 1.786 1.116 0.837 80 0.750000 7.813 6.250 3.906 3.125 1.953 1.563 0.977 0.732 90 0.666667 6.944 5.556 3.472 2.778 1.736 1.389 0.868 0.651 100 0.600000 6.250 5.000 3.125 2.500 1.563 1.250 0.781 0.586 110 0.545455 5.682 4.545 2.841 2.273 1.420 1.136 0.710 0.533 120 0.500000 5.208 4.167 2.604 2.083 1.302 1.042 0.651 0.488 130 0.461538 4.808 3.846 2.404 1.923 1.202 0.962 0.601 0.451 140 0.428571 4.464 3.571 2.232 1.786 1.116 0.893 0.558 0.419 150 0.400000 4.167 3.333 2.083 1.667 1.042 0.833 0.521 0.391 160 0.375000 3.906 3.125 1.953 1.563 0.977 0.781 0.488 0.366 170 0.352941 3.676 2.941 1.838 1.471 0.919 0.735 0.460 0.345 180 0.333333 3.472 2.778 1.736 1.389 0.868 0.694 0.434 0.326 190 0.315789 3.289 2.632 1.645 1.316 0.822 0.658 0.411 0.308 200 0.300000 3.125 2.500 1.563 1.250 0.781 0.625 0.391 0.293 210 0.285714 2.976 2.381 1.488 1.190 0.744 0.595 0.372 0.279 220 0.272727 2.841 2.273 1.420 1.136 0.710 0.568 0.355 0.266 230 0.260870 2.717 2.174 1.359 1.087 0.679 0.543 0.340 0.255 240 0.250000 2.604 2.083 1.302 1.042 0.651 0.521 0.326 0.244 250 0.240000 2.500 2.000 1.250 1.000 0.625 0.500 0.313 0.234 260 0.230769 2.404 1.923 1.202 0.962 0.601 0.481 0.300 0.225 270 0.222222 2.315 1.852 1.157 0.926 0.579 0.463 0.289 0.217 280 0.214286 2.232 1.786 1.116 0.893 0.558 0.446 0.279 0.209 290 0.206897 2.155 1.724 1.078 0.862 0.539 0.431 0.269 0.202 300 0.200000 2.083 1.667 1.042 0.833 0.521 0.417 0.260 0.195 210 0.285714 2.976 2.381 1.488 1.190 0.744 0.595 0.372 0.279 320 0.187500 1.953 1.563 0.977 0.781 0.488 0.391 0.244 0.183 330 0.181818 1.894 1.515 0.947 0.758 0.473 0.379 0.237 0.178 Because the MIDI bandwith is limited to .33ms for a one byte message, but more practically is limited to .96ms for note on messages without "intermittent" running status, all PPQNs for tempos whose 1 PPQN is less that .33ms are NOT possible and less than .96ms are NOT practical. Also, for "realistic" recording, quintuplet factorable PPQNs such as 120, 240, 240, should IMHO be avoided to prevent "false" quantization errors which I don't believe human performances ever use! Multiples of 2 and 3 are the most "natural" and basic prime quantizations. Due consideration for maximum polyphonic delays should be made. Consider the following delays REQUIRED for various simultaneous polyphony between clock ticks: Polyphony Notes at .96ms per NoteOn ========= ========================== 1 .96 ms 4 3.84 ms 8 7.68 ms 16 15.36 ms 32 30.72 ms 64 61.44 ms More expressive controller and note recordings when adequate but practical PPQNs are used to increase the efficiency of use of the sequencer's PPQN resolution. Using for example a PPQN of 1024 at a tempo of 330 BPM causes unnecessary CPU overheard and unrealizeable for MIDI's bandwith, since the 1 PPQN resolution is .178ms which is less than the .33ms that MIDI supports for transmission/reception. On the otherhand, using a PPQN of 96 for a tempo of 40 BPM prevents recording/sending of any data whose timing is less than the 1 PPQN resolution of 15.625ms! Controller events especially can "suffer", besides NoteOn event start/end times, from their recording/playback "nuance" expressiveness. Benjamin Tubb brtubb@writeme.com http://www.geocities.com/~brtubb/ http://www.pdmusic.org