On Sat, 9 Feb 2002, Tim Conrardy wrote:

> Now here is my question/observation: How does a dedicated DUMPER (like 
> Midian or Dumpit) handle the speed issue? or is it non-existent with these 
> types of applications?

I am just wondering if we are still getting mixed up here, if you read my
previous post where I quoted the Notator Manual commenting on the
background of SysEx, we are probably not rewriting against what the
experts say, we just need to understand correctly what is going on.

So in the quote above questioning how the speed for SysEx is handled, one
must accept what the experts say, which is that the dump of SysEx is not
speed controlled, this is the first mistake in our understanding on this
list, there is not anything I am proposing - that is in the tech spec that
controlls the spped in the sense we are trying to imagine, that is by
simply altering the BPM or resolution of the sequencer! Otherwise maybe if
we all in this list use hulu-hoops we might affect the rotation of the
earth as well!!

However if we read the next quote more carefully there is something to be
noticed which probably explains the so called speed dump issue at least in
this case:---

> >When I dump my friend's R70's memory into Cubase, I get a long track full
> >of midi events. 

Yes, you see, the dump is SysEx, not MIDI, but each registered event in
the sequencer has a header, the beginning of the Sysex Dump. If there are
many "MIDI" events, this is very simply many individual SysEx dumps - if
your sequencer can show the actual SysEx data, like it can in Notator,
then you can see the (often) pages of Sysex data under each Header, but
for the sake of easy-viewing, the headers are often only shown at the
time-point when the information was received.

So in the case of the R70, it is simply dumping many individual SysEx
dumps which is superficially showing the headers at each time-point they
were received.

However, for instance, on my proteus I can choose to dump the contents of
the whole sound-bank. There is a huge amount of Sysex Data, but only one
or two headers, it does not look like much unless you turn "full view" for
SysEx on. 

So the data is most certainly not cut up by the sequencer, the design is
made that it must just receive the data in one long string, until the next
break or EOX (end of message command) any new data packet gets a new
header and a new time-point in the sequencer.

The reason, I am guessing, about why the "record fast/play back slow"
advice is given is to make for most reliable operation of the process as
sequencers are all designed differently and will handle the process more
or less reliably and this advice minimises any internal problems that
sequencers can introduce into the process...

Also, the R70 will have send and receive parameters that must be met for
reliable operation and slower playback may be necessary.

Synths that have handshaking protocols have stringent timing requirements
that cannot be handled in such an inaccurate fashion as simply "playback
of data" from a sequencer, but for instance, in Notator there is provision
for you to specify the manufactures requirements for precise timing
requirements that some synths must have in order to send and receive the
SysEx data correctly.

Otherwise without this procedure, other synths have a Checksum system to
ascertain data integrity. 

I think that what is not always done on synths in their design is the
ability to simply receive the data quickly and at the full speed of the
MIDI spec, so the general advice about the speed of the sequencer

> > The manual states to record it at high speed and play back
> >at low, and the tempo does influence the amount of time it takes to dump
> >the data back to the R70.

Yes you see this would affect the speed because you are talking about many
individual SysEx dumps, as represented by the many "MIDI" events, each of
which has been received at a specific time-point, but the actual dump of
each individual event (which has actual SYsEx data under the header if
your sequencer can display it) takes place at its own speed - usually the
maximum speed opf the MIDI spec in terms of data transfer per second.

> >The only explanation I can give, reading the above, is that most
> >instruments break up a dump into small sysex messages. 

Yes, this is the case if you notice that the dump has lots of so-called
MIDI events - as this represents the individual headers of each SysEx
dump, and correspondingly a new time-point in the sequencer. It is
important to understand that, depending on the synth and the kind of dump
you initiate, you can get just one so called "MIDI" event which can have
the entire soundbank of the synth. The sequencer can only manipulate data
at a specific time-point, so even though it might take a minute or two to
receive or send such a large dump (this is not unusual), since in these
cases the SysEx data is totally represented at that time point, the
sequencer is unable to "chop" it up accordiing to its resolution, this is
an erronous understanding of the process! But if there are many individual
SysEx packets sent then each will have a new time-point and this is why it
can then be sent slow or fast depending on the BPM of the sequencer. I
also wonder if some sequencers are more suited to handling sysex than
others? For instance, in the Notator program, it is not at all necessary
to play the sequencer in order to send SysEx, if you simply click on the
header in the Event List, it will send out that particualar event. And
remember what I said in an earlier post, also in Notator, it is not
necessary to have the sequencer running in order to receive SysEx - you
see, this fact alone should make you think and realise that MIDI data and
SysEx data are different things

brian

-- 
| Brian Becroft               | Tariland, Atari user group in New Zealand |
| brian@gem.win.co.nz         | info@tariland.win.co.nz                   |
| http://gem.win.co.nz/brian/ | http://tariland.win.co.nz/                |