I recently bought an xp-50 with the idea of creating arrangements to accompany me while I play my guitar. I thought that the 20,000 note capacity was way more than I needed. I was surprised to find that with a simple four part arrangement (piano, bass, drums, organ) I ran out of memory after 40 measures. Can anyone give me some efficiency tips to better utilize the sequencer memory?

The 20,000-note memory is definitely not what the 40,000 (JV1000) or 60,000 (XP60/80) are. That said, it still holds a lot of music, especially of the kind you've described.

What you may be experiencing is that the *buffer* is full rather than the song memory itself.

Each time you perform an edit or a function, the sequencer takes a snapshop of what you already have and adds that to the buffer so you can UNDO if you need to. Once your buffer gets full, that's it.

The next time you get a memory-full message, save your tune to diskette, reinitialize your sequencer (clear its memory for a new song), and reload your tune. You can also save the tune and then turn the synth off and back on again, since the buffer is volatile. That will clear the buffer/memory.

The problem I had with the XP50 (before I got an XP60) was that I had big (dense) songs -- which the XP50 held just fine -- but couldn't do big global edits because the buffer got full so fast for UNDO. The bigger the area you edit and is held in the UNDO buffer, the less memory is left.

See if this helps.

I also used to have that problem. I would write songs that would contain 120-140 measures at a time. The buffer would definitely get full as just described. I had to save songs to disk and then intialize to reclaim space. Sometimes, the DATA THIN function works. Be warned however. The DATA THIN function may also not work id the data buffer is full as well.

Getting an XP60 helped alleviate this problem for the most part. The increased memory really helped. The fact that I write HUGE arrangements is why in fact.

The Infamous Epu.

Thanks for the tip. I will try it tonight. Do you know of a way to disable the UNDO feature to prevent data being copied to the buffer? I know this is risky in that you wouldn't be able to undo an inferior variation. However I would be willing to take this risk to avoid the "memory full" problem.

Thanks for the suggestion. I will try the reinitializing tip. Could I create patterns (sections of a song) and link them together as a work-around for memory limitation? I'm not sure if you can link patterns together that are not part of the same song though. I downloaded some classical arrangements that are much larger than what I'm trying to create, and I am puzzled by how such a huge file could be handled by the xp-50. Could it be that they are breaking the composition into smaller pieces?

Well, 1 thing is certain, you aren't using up 20,000 notes in 40 measures. That'd be 500 notes per measure!
One question of curiosity. Does the number of tones per patch or part have anything to do with notes? For instance, if you run a part that uses all four layered tones, is that taking 4 notes per key you press on the sequencer? I know it works that way with polyphony, does it work that way with the sequencer?

Hei everyone.

First: No, the number of tones in each patch has nothing to do with the capacity used in the sequencer. The sequencer just sends a note over the correct MIDI-channel, and it is up to the synth-section of the XP to actually produce the sound. It does not matter to the sequencer if the sound produced is simple or complex.

Another thing to remember is that the sequencer stores 20000 MIDI-events, and not notes. This means that if you hava a lot of controller-data, sysex, program changes etc, this will use up the sequencer capacity. One prime example here is the aftertouch. If you press a key and make a long sustained note while changing the aftertouch pressure, the sequencer will only record one note-event, but perhaps hundreds of Control-Change-events for the aftertouch.

Data thin helps. It reduces the amount of controller data. This will not normally be audible, but the sequencer capacity is extended. If you are having problems with sequencer-memory, try to use data thin on each track after you have recorded them.

Another trick is to switch off aftertouch on parts where you don't need it, so that you are sure that no aftertouch-changes are generated.

Hope this was understandable.

Stig

Good point about the controller, PB, etc. data. If you don't watch out, it can use up gobs of memory in a hurry.

Rather than data thin, I think probably he's in a position just to use . . . ERASE That is, he just needs to erase *all* the controller data on *all* the tracks where such data has been *inadvertently* entered. I do that periodically during every recording session. Or, as you say, turn it off entirely.

As for patterns: "Could I create patterns (sections of a song) and link them together as a work-around for memory limitation? I'm not sure if you can link patterns together that are not part of the same song though"

I use patterns all the time, all over my songs for precisely this reason: If something repeats, the pattern message takes up much less memory than pasting the whole measure over again.

Yes, you can paste patterns together from different songs . . . sorta. You have to load them individually from disk from your different songs and then string them together (with pattern call messages) in the internal song. You must then save that internal song + the imported patterns into a single song.

Is any of this helping? You know, if you're still having trouble, you could probably send your .svq file to some of us on the forum here to have a look at.

Anyway, good luck. You should have plenty of room in the XP50 for what you're trying to do.

Thanks for the advice. I tried your suggestion from yesterday with the same result. Actually I didn't even get to edit the song. I tried to insert two measures after #34 and got the memory full message.

I don't think I used any controllers while recording, but I will try to delete them if I did.
I saved my song to a SMF file and tried to edit it in cakewalk (my puter is in a different room) and noticed that the piano track size was over 10,000. Assuming that size and notes are approximately the same, some controller data must be in the track, since I basically just played a 4 note chord and a few arppregios per measure.

I'll keep tinkering with it until I figure it out. Otherwise I will submit the svq file as suggested and solicit your comments.

Your answer re. patterns causes this question. Are patterns then stored internally? If not, are they stored on disk separate from the song?

Thanks for all the help.

Thanks for your reply. Your mathematical analysis indicate that there is some other problem. I am a poor piano player, and mostly just hit one 4-note chord per measure. Using cakewalk I noticed that the size of the piano track was over 10,000. If these are all notes, then that is over 250 per measure. There is no way I could play that many even if I wanted to. I will try to determine if there is controller data that were inadvertenly included in my sequence. Thanks for the help.