[sdiy] USAMO

[Ove Ridé]

On 26 March 2015 at 15:12, Nantonos <nantonos at epona.net> wrote:
> Hello Adam,
>
> Thursday, March 26, 2015, 7:36:43 AM, you wrote:
>
>> http://expert-sleepers.co.uk/usamo.html
>
>> Anyone seen this? This looks very clever.
>> Sample accurate midi generated from an audio track - not just clock (a
>> la the Innerclock Sync Gen, et al)  - but ALL midi messages.
>
> That page says
>
>   "Because the MIDI signals are generated and transported as audio,
>   they are guaranteed to be sample-accurately synced with your audio,
>   and free of the jitter often associated with computer-generated
>   MIDI."
>
> this seems to be an inaccurate claim. It would be possible to do this,
> but the same box would need to either have several channels of audio
> in and out (plus preamps, etc) to be sample-accurate *with*, or
> alternatively some digital in-out, for example ADAT, plus a wordclock
> source so that some other box could be kept aligned to a granularity
> of one sample.
>
> As a standalone box, its not sample-accurate with anything else.

I think you are perhaps misreading the claim? In the simplest setup,
you have an audio interface with one audio channel outputting audio,
and one audio channel outputting encoded MIDI to this box. Assuming
the channels of your audio interface are locked to each other, you can
predictably start an audio sample and a MIDI message at the same
sample position, on the two different channels. You could also start
the MIDI message one sample position later etc. You would still have a
decoding delay, but if this box is well designed, this delay will be
constant and can be compensated by sending the MIDI message earlier if
you need/want such precision. You would only need a wordclock if you
wanted to sync multiple DACs/ADCs so the sample input/output doesn't
drift between individual DACs/ADCs. This box is likely receiving a
message in an asynchronous format, in the simplest form UART over an
audio channel. It doesn't need to be aware of the samplerate of the
transmitting audio interface, but that doesn't stop you from nudging
the start of the message back or forward by individual samples.

-- 
/Ove

Blog: <http://blog.gg8.se/>

"Here is Evergreen City. Evergreen is the color of green forever."