Yahoo Groups archive

Disklavier

Index last updated: 2026-04-28 23:20 UTC

Message

Re: MIDI Velocities

2005-09-01 by George F. Litterst


Good morning, everyone.

On Sep 1, 2005, at 2:28 AM, Bob wrote:
Hello, George
Yes, I am using a computer to play the DLV through the MIDI In/Out plugs with the 500 ms delay OFF. I turned it off so that the keyboard graphic on the karaoke player would be in sync with the real piano keys.

Good to hear that. This explains the trouble that you were having. The better solution is to play the MIDI files from the computer with the Disklavier delay turned on. If you want karaoke lyrics to be in sync with the piano, open the MIDI file in a sequencer, select all of the lyric data, and shift the lyric data toward the back of the file by 500 ms.

I am willing to play the same files that dropped notes with the delay on. I just didn't understand your explanation. You spoke of the "note off" coming before the key actually struck with the delay off. Thus causing a note drop. But if the delay was in place as you suggest, isn't it true that the very same things would happen just 500 ms later?

Consider a simple MIDI situation:

You are sending MIDI data from a computer to an electronic device (such as a digital piano). For all practical purposes, the MIDI data arrives instantly at the playback device. Because the playback device is electronic, it responds instantly, and you hear the result instantly.

(I do realize that there are some tiny latencies in this scenario, but let's forget that for the sake of argument.)

When you send messages to play notes, you actually send 2 messages per note: a note-on message and a note-off message.

Now, let's compare this scenario to your situation in which you are sending MIDI data to the Disklavier:

Your computer sends the MIDI data and it arrives instantly at the Disklavier. If the Disklavier's delay is turned off, the Disklavier responds instantly (just like the digital piano responds instantly). HOWEVER, because the Disklavier is a mechanical instrument, it responds instantly not by making a sound that you hear immediately but by moving the keys. It takes (I think) anywhere from 30-90 ms for the hammers to hit the strings.

Exactly how long it takes for the hammers to hit the strings depends upon the velocity parameter of the note-on message. Louder notes will be heard a bit sooner than softer notes. As a result, the timing is a little sloppy compared to the original MIDI file.

Now, imagine that you are sending messages for a rapidly repeating note: note-on; note-off; note-on; note-off; etc. It is entirely possible that some of your note-off messages will interrupt the key stroke of the previous note-on message and result in weak-sounding or non-playing notes.

If you engage the Disklavier's delay feature, the Disklavier will buffer all incoming data by 500ms. During that 500ms delay, the Disklavier will examine the incoming data, determine the desired results, and compute appropriate start and stop times for each key stroke so that the desired results will be achieved. In other words, the 500ms delay will take care of the problem in which loud notes sound sooner than soft notes and it will avoid the problem of rapidly repeating notes being dropped.

NOTE: The typical Disklavier (especially the Pro) can outplay most pianists in terms of speed. However, it is possible to create a MIDI data stream on a computer that cannot be played by a mechanical instrument (even though it can be played by a purely electronic instrument).

Regards,
PianoBench

Attachments

Move to quarantaine

This moves the raw source file on disk only. The archive index is not changed automatically, so you still need to run a manual refresh afterward.