Subject: Re: [linux-audio-dev] Utility for recording audio track with midi clock
From: Jens M Andreasen (jens.andreasen_AT_chello.se)
Date: Fri Jul 02 2004 - 13:04:15 EEST
On fre, 2004-07-02 at 10:20, Alfons Adriaensen wrote:
> On Fri, Jul 02, 2004 at 07:48:29AM +0200, Jens M Andreasen wrote:
>
> > FWIW: My experience with external sequencers is that they'll output all
> > notes due up to the most recently received time, which is also what is
> > expected.
>
> IF that is all that's required then sending MIDI clock from a 1000 Hz RTC
> should work perfectly. The added jitter of +/- 0.5 ms is less than the time
> it takes to transmit a 3-note chord (2.24 ms, with running status) from
> a keyboard (assuming someone can play that accurately).
>
> But iff the receiver wants to compensate for the transmission delay, it
> will need to have some idea of the 'current speed'.
Midi time (the quarter note varaint) has no idea of "current speed".
Your sequencer probably shows a measure of BPM, but this is not
reflected anywhere in the protocol. Also, the time resolution is really
bad. You can't use it for jazzy realtime impressions where *exact*
timing matters (unless you pull up BPM well beyond 240, but then you'll
loose bandwith ...).
If you want to interpolate between clock events, then midi-smpte is the
only reliable route to go.
Keeping arpeggiators (etc) in sync with each other is better done with
midi-time though.
To clarify:
midi-time is your (somewhat whimsical?) conductor.
midi-smpte is the wall clock (on some wall.)
mvh // Jens M Andreasen
This archive was generated by hypermail 2b28 : Fri Jul 02 2004 - 12:52:23 EEST