Reference Manual

584 MIDI Fact Sheet
3. Playthrough involves sending MIDI note and controller information from a hardware de-
vice (such as a MIDI keyboard) into the DAW and then, in real-time, back out to a hard-
ware synthesizer or to a plug-in device within the DAW. An ideal playthrough environment
would “feel“ as accurate and responsive as a physical instrument such as a piano.
34.2 MIDI Timing Problems
The realities of computer-based MIDI are complex, and involve so many variables that the ideal
systems described above are impossible to achieve. There are two fundamental issues:
1. Latency refers to inherent and consistent delay in a system. This is a particular problem in a
DAW because digital audio cannot be transferred into or out of an audio interface in real
time, and must instead be buffered. But even acoustic instruments exhibit a certain degree
of latency; in a piano, for example, there is some amount of delay between the time a key
is depressed and the time the hammer mechanism actually activates the string. From a per-
formance perspective, small latency times are generally not a problem because players
are usually able to adapt the timing of their playing to compensate for delays — as long as
the delays remain consistent.
2. Jitter refers to inconsistent or random delay in a system. Within a DAW, this can be a par-
ticular problem because different functions within the system (e.g., MIDI, audio and the
user interface) are processed separately. Information often needs to be moved from one
such process to another — when converting MIDI data into a plug-in‘s playback, for ex-
ample. Jitter-free MIDI timing involves accurate conversion between different clocks within
the system‘s components — the MIDI interface, audio interface, and the DAW itself. The
accuracy of this conversion depends on a variety of factors, including the operating sys-
tem and driver architecture used. Jitter, much more so than latency, creates the feeling that
MIDI timing is “sloppy“ or “loose.“
34.3 Live’s MIDI Solutions
Ableton‘s approach to MIDI timing is based on two key assumptions:
1. In all cases, latency is preferable to jitter. Because latency is consistent and predictable, it
can be dealt with much more easily by both computers and people.