Disklavier

Good morning, everyone.

In a message dated 9/28/02 10:57:34 PM, Carol writes:

<< BTW, where does one look for that meta tag that assigns the piano parts?>>

I was afraid that someone was going to ask that. First, some background:

Most computer sequencers are not true MIDI editors. They open and save files 
in their own format. True, they do store MIDI data and should comply with the 
official MIDI spec. Nonetheless, they have their own format which allows them 
to let you save all sorts of preferences that apply only to your particular 
situation (like the layout of the  screens on your computer, et cetera) and 
may include data types that have nothing to do with MIDI (such as audio 
tracks).

If you use a typical computer sequencing program to edit a SMF, the program 
actually imports the SMF into its own format, lets you do the editing, and 
then exports the data back to SMF. During the conversion to and from SMF, the 
sequencer may choose to reorder some of the MIDI events in a subtle way (and 
that normally doesn't matter, unless you are editing XP data from a 
Disklavier Pro) and the sequencer may choose to ignore and discard certain 
MIDI messages.

For reasons that I cannot fathom, nearly every computer sequencer that I have 
encountered discards a type of MIDI message called a "sequencer-specific meta 
event." That is the type of event that we are discussing.

A sequencer-specific meta event is similar to a system exclusive (SysEx) 
message.  It begins with a manufacturer's ID and then continues with other 
hexadecimal numbers. I am not sure what the technical distinction is between 
the intended functionality of SysEx and sequencer-specific meta events, but I 
suspect that the latter is capable of holding a much longer string of numbers.

Most sequencers do save the SysEx messages.

The MIDI Manufacturers Association encourages companies to publish their 
manufacturer-specific messages so that everyone can take advantage of them. 
Many of these messages, however, are undocumented.

The fact that these messages begin with a manufacturer's ID means that it is 
easy for devices that do not understand these messages to ignore them.

There are only two computer sequencing programs with which I am familiar that 
display and save sequencer-specific meta events: Yamaha's XGWorks and 
Tontata's MIDIGraphy (http://member.nifty.ne.jp/mmaeda/e/works.html). The 
former is available for both Mac and Windows (but I can only confirm that the 
Windows version saves these meta events; I presume that the Mac version does 
as well). MIDIGraphy is for the Mac only and is a true, native MIDI file 
editor.

When you make a recording on a Disklavier, model Mark IIXG or later, the 
Disklavier puts this meta event into the file as a default:

43 7B 0C 02 01

(Note: This is hexadecimal. The 0s are zeros, not capital letters.)

"43" is Yamaha's ID code. I presume that "7B 0C" identifies the purpose of 
the meta event. "02" tells the Disklavier to play the data on channel 2 on 
the piano and assign that data to the R Part Cancel button. "01" tells the 
Disklavier to play the data on channel 1 on the piano and assign that data to 
the L Part Cancel button.

(As an aside, I'll mention that the Yamaha CVP series Clavinovas makes use of 
these messages, too. In the absence of these messages, the Clavinova would 
assign data on channel 1 to R and data on channel 2 to L. The Disklavier does 
the opposite. The presence of these messages insures that both types of 
instruments read the data as it was intended by the author of the file.)

The publishers of pedagogical MIDI files that coordinate with piano teaching 
methods (such as those from Hal Leonard, Alfred, Warner Bros., FJH, et 
cetera) have been putting this version of the meta event into their files in 
recent years:

43 7B 0C 04 03

As you can see, this message tells Yamaha instruments to assign data on 
channel 4 to R and  data on channel 3 to L.

You can use your own Mark IIXG Disklavier or later model to add these meta 
events to a Type 0 SMF. (If you have a Type 1 SMF, I think you will have to 
convert it to Type 0 first, which you can also do on the Disklavier.) To add  
this event:

--Press Record
--Use the backward Song Select button to choose the SMF in question
--Press the Right Arrow button repeatedly to get to the last screen, where 
you will see L= and R=

If the file already has the meta event, you will be able to see what L and R 
are set to. If the file does not have the meta event, it will appear as 
though L=1 and R=2. This is not actually the case!

At this point, you must use the + and - buttons to change the values for L 
and R. (Even if you want the message to be L=1 and R=2, use these buttons to 
change the default values and then to set them back to 1 and 2.) When you are 
done, press Stop and then press Enter to save your changes. Your file now as 
this meta event.

<<How do you recommend reducing those "note-on" velocities - edit them 
individually or globally? >>

I have not used the Velocet program. I know that others on this list have 
suggested ways of reducing velocities in a way that presumably preserves the a
 proper scaling for the velocities. I personally take a much more cavalier 
approach. I first set controller 7=100 for the piano tracks. Then, if they 
are louder than I like, I select all of the piano notes at once and subtract 
some number of velocity points from them, such as some number between 5 and 
30. Based on experience or by glancing at the velocities in the file, I can 
usually guess the right amount on the first try. However, trial-and-error do 
not take very long.

In general, my feeling is that most files that were not recorded on a real 
piano are a bit artificial, as far as note-on velocities are concerned. If 
they were recorded on a digital piano, for example, the touch of the pianist 
would have been heavily influenced by the setting of the speaker volume.  Who 
knows where that was set. Given the artificial circumstances of the original 
recording, I don't worry too much about maintaining an authentic scaling of 
the velocities for Disklavier. I am not saying that my opinion is any better 
than anyone else's. It is simply my opinion.

Hope that helps.

Regard,
PianoBench