On Mon, 2010-06-21 at 11:39 +0100, Daniel James wrote:
> Hi Gustin,
> >> Also, we have less on-board MIDI controllers now, and more USB, which I
> >> suspect might be contributing to the problem.
> > Is there some way to test this?
> For ALSA MIDI there is: http://github.com/koppi/alsa-midi-latency-test
> Ralf has been using some of Fon's tools; you'll have to ask him about
> the details of his methodology.
> > You are not going to see 1 ms timing with USB2 gear.
> That seems to be the case, but then USB is on every system and MIDI
> sockets are not, which has skewed the keyboard and controller market
> away from better technical solutions.
> Even RME have two USB MIDI interfaces now (Fireface UC and Babyface). It
> just makes me glad I kept my five-pin DIN MIDI cables :-)
I'm not sure if it's worth the effort. Perhaps I got 10 mails off-list
confirming this issue and another 10 mails off-list with name-calling
because it's again FUD and that there are millions of anonymous people
who ask to ban me from all Linux audio lists. So I guess I unsubscribe
to all Linux audio lists myself and repair my Atari ST.
Anyway, how to use the Fon's tool. Connect your hw MIDI out with your hw
MIDI in, that's it.
My tests were done by recording beats from external MIDI equipment, e.g.
a very short sinus (an impulse) from a DX7 at 120 BPM.
Then I zooooooooooooomed into the waveforms and searched for the
beginning of the waveforms, that's very hard work, because noise might
cover the start points.
120 BPM = every 500ms one beat, but recorded beats might be at 448ms or
502ms = - or + 2ms. I recorded many beats to see what's the max and
what's the min jitter.
Because an audio track might have latency and we can set offsets I did
max_jitter - min_jitter = effective_jitter
64studio-users mailing list