MakeMusic
SmartMusic Finale Garritan MusicXML

forward and backup

Moderator: Michael Good

forward and backup

Postby Guido Amoruso » Wed Nov 06, 2002 12:53 pm

Hi,

I'm rather new to MusicXML and I'm sorry if this is a trivial question, but I cannot figure out this matter without any help. I'm trying to understand the way MusicXML deals with polyphony: I've read the thread about the backup element and got some hints, but I still cannot understand these things:

1. what is the need for the forward element? Why is not enough the backup element? That is, I see that some times there is a backup till the start of the measure and then a forward to some time onwards, but wouldn't be the backup enough?

2. after understanding the preceding matter, which ways the "Forward is generally used within voices and staves, while backup to move between voices and staves."? <from the note.dtd>

Thank you very much for any hints!

Guido Amoruso
Guido Amoruso
 
Posts: 10
Joined: March, 2014
Reputation: 0

RE: forward and backup

Postby Michael Good » Wed Nov 06, 2002 2:56 pm

Hi Guido,

This is not a trivial question at all, and I see that our sample files at www.musicxml.org/xml/samples.html did not really cover this case. So I have added a new example (from Brahms's song "Wie Melodien zieht es mir", Op. 105, No. 1) for this case.

Look at measure 1 in the piano part in this song. The right hand is cross-staved into the left hand part on beat 4, so the beamed group is all represented in voice 1. The piano part then backs up to the half note and quarter rest in the right hand. But now what? There's an extra beat but no quarter rest. We want to represent that extra beat so the timing works out properly for MIDI applications. Therefore, we put in a <forward> to fill out the full measure in the left hand with an invisible rest.

This is a typical case where the forward is used within a staff or voice, while the backup is used in moving between staves or voices.

You are correct that when a voice starts in the middle of a bar, you could just use a <backup> instead of a <backup> and <forward>. But for some software, it's easier to generate the two elements instead of combining them.

The most obvious example for needing a <forward> element is where voices disappear in the middle of a bar. However I didn't find an example in my quick perusal of our song files. If I find it later, I will add that to our XML samples.

Please let me know if you have further questions after looking at this sample.

I'll also add a sample of multiple slurs soon. We have some a good example of overlapping slurs used for different versions of a repeat in one of the songs we are about to add to our store. Once the new songs are online, I will add this to our sample collection.

Best regards, Michael
Michael Good
VP of MusicXML Technologies
MakeMusic, Inc.
User avatar
Michael Good
 
Posts: 2197
Joined: March, 2014
Reputation: 0

RE: forward and backup

Postby Guido Amoruso » Wed Nov 06, 2002 5:17 pm

On Wed, 2002-11-06 at 21:56, Michael Good wrote:
Michael Good wrote:Therefore, we put in a <forward> to fill out the full measure in the left hand with an invisible rest. So, you say, there is one and only "time stream" for each voice in a bar, which can start not at the beginning of the measure, but once started, it has to go on without breaks till the end (through each

"timed" note or maybe some silent forward), except for meeting a <backup> which hangs up this voice instance. And we can have more than one voice instance. Is that correct? (I now it's a twisted summary!) I mean, for example, I could not have a quarter note belonging to another voice (yes, it makes no sense, it's just an example!) instead of the last <forward> in the first measure of your Brahms example: if I wanted it I should backup [at the beginning] and than go forward again.

Michael Good wrote:You are correct that when a voice starts in the middle of a bar, you could just use a <backup> instead of a <backup> and <forward>. But for some software, it's easier to generate the two elements instead of combining them. I have noticed in some samples (Dichterliebe) that there are multiple <forward>

elements containing a <duration> matching that of real notes (one <forward> corresponding to a quarter note plus one corresponding to a quaver, for example): should I assume that this is a rule for <forward> or not? It could be useful for me to know that. If so, than it could be useful to have another rule, too, for backup elements leading always at the beginning of the measure, so I could be guided till the new voice beginning by a path of silent rests. I don't know whether this makes sense or not, but you said the for some software does, so I'm wondering if this can be a general behaviour.

Thank you very much. Regards, Guido
Guido Amoruso
 
Posts: 10
Joined: March, 2014
Reputation: 0

RE: forward and backup

Postby Michael Good » Wed Nov 06, 2002 6:20 pm

Hello Guido,

No, there's only one time stream in a MusicXML part; not one time stream per voice.

Our examples all come from running our Dolet for Finale converter, so they reflect how Finale's software works. For Finale, it is easiest to go one layer at a time, and use separate MusicXML <voice> elements for each layer and voice in Finale. But other software does not work this way, and there is no need to follow Finale's example if it does not make sense.

Yes, you could replace the <forward> in this example with a quarter note in another voice. You would not need the <backward> and <forward>
elements, and it would probably be "nicer" if you did not include them. But if it's too hard to do it that way, don't worry about it.

Similarly, the examples have multiple <forward> elements when that is how things are represented in Finale. If there is an invisible quarter rest followed by an invisible eighth rest, it will be represented as two <forward> elements, rather than a single dotted quarter <forward>
element. Why do extra work that is not needed?

It is good if MusicXML voices represent actual musical voices, but this is not always feasible. MuseData files tend to do a very good job of voice assignment, so those files usually reflect voice leading very accurately. Converting some MuseData files to MusicXML using the full version of Dolet will give you some different examples of MusicXML style.

On the other hand, most Finale engravers will use layers and voices based on what makes it easiest to get the correct appearance on the page, not based on an analysis of voice leading. So the results will be mixed, unless you let them know that voice leading matters to you.

Does that make things more clear? Or more confusing?

Best regards, Michael
Michael Good
VP of MusicXML Technologies
MakeMusic, Inc.
User avatar
Michael Good
 
Posts: 2197
Joined: March, 2014
Reputation: 0

RE: forward and backup

Postby Guido Amoruso » Thu Nov 07, 2002 2:31 am

> Does that make things more clear? Or more confusing? Much more clear, thank you! I'll make you know whether I have further questions in the future.

Regards, Guido
Guido Amoruso
 
Posts: 10
Joined: March, 2014
Reputation: 0

Who is online

Users browsing this forum: No registered users and 2 guests

cron