MakeMusic
SmartMusic Finale Garritan MusicXML

Sibelius Standard SoundWorld (S3W) as a possible means of describing sounds in MusicXML 3.0

Moderator: Michael Good

Sibelius Standard SoundWorld (S3W) as a possible means of describing sounds in MusicXML 3.0

Postby Daniel Spreadbury » Fri Jan 28, 2011 9:31 am

Hi everyone,

With reference to one of Michael's stated goals for MusicXML 3.0:

> - Add a standard instrument taxonomy to indicate that a given
> instrument in a score should sound like a trumpet, guitar, piano, etc.

I would like to propose that the taxonomy used by Sibelius 5 and later, known as the Sibelius Standard SoundWorld (hereafter S3W) is considered for use in MusicXML 3.0. If the community decides that it is suitable, we will look into releasing S3W under a royalty-free license similar in style to that used by MusicXML itself in order to allow everybody who is working with MusicXML to make use of the taxonomy.

It's worth pointing out up-front that SoundWorld is specifically a taxonomy of instrument *sounds* rather than instruments themselves, and I believe this has specific benefits that make it especially suitable for solving the above-stated goal.

I have made available a paper on SoundWorld written by Jonathan Finn, who along with me and a couple of other people here conceived the SoundWorld system. It's quite a lengthy read, but it provides details about the construction of sound IDs, organisation into sound worlds, how the substitution rule to use in the event of a specified sound not being available should work, how some sound ID elements are mutually exclusive, and so on.

The paper can be downloaded here (1.9MB, PDF):

http://www.sibelius.com/download/SoundWorld.pdf

All the best,

Daniel

-- Daniel Spreadbury Senior Product Manager | Sibelius R & D

Avid Unit 20-23 City North Fonthill Road London N4 3HF United Kingdom

t +44 (0)20 7561 7915 | m +44 (0)7990 558997 | f +44 (0)20 7561 7888

Sibelius is Avid. Learn more at www.avid.com <http://www.avid.com/>

Avid Technology Europe Limited is a limited company registered in England and Wales. Registered No.: 2670844. Registered Office: Pinewood Studios, Pinewood Road, Iver Heath, Bucks SL0 0NH, England
Daniel Spreadbury, Steinberg
http://blog.steinberg.net
User avatar
Daniel Spreadbury
 
Posts: 13
Joined: March, 2014
Location: London, UK
Reputation: 0

Re: Sibelius Standard SoundWorld (S3W) as a possible means of describing sounds in MusicXML 3.0

Postby Joe Berkovitz » Tue Feb 01, 2011 5:35 pm

Daniel,

Thanks for generously making the SoundWorld/S3W standard available for consideration here.

I think that SoundWorld has many aspects that are helpful for this aspect of MusicXML. The exact syntax and organization of the S3W taxonomy is not so important from my point of view (and I have no particular issues with it either); the main concepts I'd like to see imported from S3W are the overall structure and the approach to substitution. I think it's a good starting point for thinking about what would work in MusicXML 3.0.

For me, the main question to be examined in SoundWorld is this:

Should sound IDs really encode things like staccato, marcato, etc? I find this problematic since it overloads the notion of an "instrument", and articulations are already covered elsewhere in MusicXML. I believe what the MusicXML spec is looking for is something more like InstrumentWorld ;) An instrument produces a family of timbres as dictated by performance directions
-- it is not just one timbre. Of course, one may have an instrument that is played in a particular technique built around one specific articulation -- say, an entire flute part played in flutter-tongue. That's not problematic since it's in effect an exotic instrument sound, not an articulatory variant.

So what I am concerned about is a direction where specific articulations or performance techniques on individual notes wind up getting mapped to different sound IDs. That feels too granular to me and takes us away from the idea of an
"instrument". It is likely to confuse implementers and lead to conflicting interpretations. Even the idea of ensemble vs. solo instruments is questionable -- couldn't these be variants, automatically selected by some programs to render performance directions identifying specific solo passages?

I did read the defense of this idea in the spec, but I didn't find myself buying into it at the end. It feels like a good way to organize sounds, but overly fine-grained for instruments. I suppose this is restating exactly what you pointed out -- SoundWorld is a sound taxonomy, not an instrument taxonomy. I guess I have some discomfort with that, while liking the rest of the concept quite a lot.

Best,

... . . . Joe

Joe Berkovitz President Noteflight LLC 84 Hamilton St, Cambridge, MA 02139 phone: +1 978 314 6271 www.noteflight.com


Daniel Spreadbury wrote:Hi everyone,

With reference to one of Michael's stated goals for MusicXML 3.0:

Michael Good wrote:- Add a standard instrument taxonomy to indicate that a given instrument in a score should sound like a trumpet, guitar, piano, etc.

I would like to propose that the taxonomy used by Sibelius 5 and later, known as the Sibelius Standard SoundWorld (hereafter S3W) is considered for use in MusicXML 3.0. If the community decides that it is suitable, we will look into releasing S3W under a royalty-free license similar in style to that used by MusicXML itself in order to allow everybody who is working with MusicXML to make use of the taxonomy.

It's worth pointing out up-front that SoundWorld is specifically a taxonomy of instrument *sounds* rather than instruments themselves, and I believe this has specific benefits that make it especially suitable for solving the above-stated goal.

I have made available a paper on SoundWorld written by Jonathan Finn, who along with me and a couple of other people here conceived the SoundWorld system. It's quite a lengthy read, but it provides details about the construction of sound IDs, organisation into sound worlds, how the substitution rule to use in the event of a specified sound not being available should work, how some sound ID elements are mutually exclusive, and so on.

The paper can be downloaded here (1.9MB, PDF):

http://www.sibelius.com/download/SoundWorld.pdf

All the best,

Daniel

-- Daniel Spreadbury Senior Product Manager | Sibelius R & D

Avid Unit 20-23 City North Fonthill Road London N4 3HF United Kingdom
>
t +44 (0)20 7561 7915 | m +44 (0)7990 558997 | f +44 (0)20 7561 7888

Sibelius is Avid. Learn more at www.avid.com <http://www.avid.com/>

Avid Technology Europe Limited is a limited company registered in England and Wales. Registered No.: 2670844. Registered Office: Pinewood Studios, Pinewood Road, Iver Heath, Bucks SL0 0NH, England
Joe Berkovitz
 
Posts: 79
Joined: March, 2014
Reputation: 0

Re: Sibelius Standard SoundWorld (S3W) as a possible means of describing sounds in MusicXML 3.0

Postby Daniel Spreadbury » Tue Feb 01, 2011 6:10 pm

Joe Berkovitz wrote:For me, the main question to be examined in SoundWorld is this:

Should sound IDs really encode things like staccato, marcato, etc?

For Sibelius's purposes, absolutely they should. This is because sounds in those playback devices (sample libraries et al) that S3W is designed to abstract away from are often organised in those terms. Sibelius needs to be able to find the sound ID used at any point along a given staff so that its fallback and substitution rules can be applied not only to the initial sound used by an instrument, but also to any and all sound changes (typically technique changes such as staccato or marcato) required by the playback of that staff.

However, if your question is whether sound IDs *in MusicXML* should encode things like staccato, marcato, etc., then I think the answer is different. Just as in Sibelius you would rarely define, say, strings.violin.ensemble.staccato as the default (initial) sound ID for an instrument, you would most likely not want to do so in MusicXML either.

But just because you might not want to include effects in the sound ID encoded for a given instrument doesn't mean, IMO, that their use should be precluded
(indeed, you might well end up using the "pizzicato" effect legitimately in an initial sound ID that describes the sound made by an upright bass).

Likewise, it may be useful for some applications that consume MusicXML files to have optional information available about e.g. a text direction such as
"sul pont." in order to provide an appropriate change of playback sound at that point. So being able to define a sound ID *change* (which would typically be the addition or subtraction of one or more effects) may well be useful for some applications, particularly if their needs are slightly more sophisticated than simply choosing a more appropriate default sound for a given part in the MusicXML file.

Daniel

-- Daniel Spreadbury Senior Product Manager | Sibelius R & D

Avid Unit 20-23 City North Fonthill Road London N4 3HF United Kingdom

t +44 (0)20 7561 7915 | m +44 (0)7990 558997 | f +44 (0)20 7561 7888

Sibelius is Avid. Learn more at www.avid.com <http://www.avid.com/>

Avid Technology Europe Limited is a limited company registered in England and Wales. Registered No.: 2670844. Registered Office: Pinewood Studios, Pinewood Road, Iver Heath, Bucks SL0 0NH, England
Daniel Spreadbury, Steinberg
http://blog.steinberg.net
User avatar
Daniel Spreadbury
 
Posts: 13
Joined: March, 2014
Location: London, UK
Reputation: 0

Re: Sibelius Standard SoundWorld (S3W) as a possible means of describing sounds in MusicXML 3.0

Postby Joe Berkovitz » Tue Feb 01, 2011 6:28 pm

Hi Daniel,

I think we're in agreement here (hopefully non-violent agreement ;) Yes, I only meant sound IDs in MusicXML -- I understand how sound IDs work in Sibelius and agree that they serve the right purpose for the approach that was taken there. I'd still argue that syntactically distinguishing effect nodes from instrument nodes is a good idea, but that's irrelevant to this discussion.

I also agree that there could be some hypothetical mechanism which maps performance directions and effects to specific sounds in a library. However, it is sometimes the case that such effects are directly implemented by a program's synthesis module, rather than being distinct sounds in themselves -- in which case there may not be a need for such a mapping. So this feels to me like an optional project to take on at the moment.

...joe


Daniel Spreadbury wrote:
Joe Berkovitz wrote:For me, the main question to be examined in SoundWorld is this:

Should sound IDs really encode things like staccato, marcato, etc?

For Sibelius's purposes, absolutely they should. This is because sounds in those playback devices (sample libraries et al) that S3W is designed to abstract away from are often organised in those terms. Sibelius needs to be able to find the sound ID used at any point along a given staff so that its fallback and substitution rules can be applied not only to the initial sound used by an instrument, but also to any and all sound changes (typically technique changes such as staccato or marcato) required by the playback of that staff.

However, if your question is whether sound IDs *in MusicXML* should encode things like staccato, marcato, etc., then I think the answer is different. Just as in Sibelius you would rarely define, say, strings.violin.ensemble.staccato as the default (initial) sound ID for an instrument, you would most likely not want to do so in MusicXML either.

But just because you might not want to include effects in the sound ID encoded for a given instrument doesn't mean, IMO, that their use should be precluded (indeed, you might well end up using the "pizzicato" effect legitimately in an initial sound ID that describes the sound made by an upright bass).

Likewise, it may be useful for some applications that consume MusicXML files to have optional information available about e.g. a text direction such as
"sul pont." in order to provide an appropriate change of playback sound at that point. So being able to define a sound ID *change* (which would typically be the addition or subtraction of one or more effects) may well be useful for some applications, particularly if their needs are slightly more sophisticated than simply choosing a more appropriate default sound for a given part in the MusicXML file.

Daniel

-- Daniel Spreadbury Senior Product Manager | Sibelius R & D

Avid Unit 20-23 City North Fonthill Road London N4 3HF United Kingdom
>
t +44 (0)20 7561 7915 | m +44 (0)7990 558997 | f +44 (0)20 7561 7888

Sibelius is Avid. Learn more at www.avid.com <http://www.avid.com/>

Avid Technology Europe Limited is a limited company registered in England and Wales. Registered No.: 2670844. Registered Office: Pinewood Studios, Pinewood Road, Iver Heath, Bucks SL0 0NH, England

... . . . Joe

Joe Berkovitz President Noteflight LLC 84 Hamilton St, Cambridge, MA 02139 phone: +1 978 314 6271 www.noteflight.com
Joe Berkovitz
 
Posts: 79
Joined: March, 2014
Reputation: 0

Re: Sibelius Standard SoundWorld (S3W) as a possible means of describing sounds in MusicXML 3.0

Postby Hartmut Lemmel » Wed Feb 02, 2011 6:00 am

Hi Joe, it seem that the capella genericSound IDs would exactly meet your wishes.

See http://lemmel.at/captune-VST-config.pdf page 5ff.

It firstly defines the instrument and then, optionally, the articulations. (In capella we don't substitute articulation sounds, but only instruments.) I think this structure would also fit the providers of sound libraries, as they usually also classify their sounds in two levels, first by instrument, then by articulation.

Best regards, Hartmut

Joe Berkovitz wrote:Daniel,

Thanks for generously making the SoundWorld/S3W standard available for consideration here.

I think that SoundWorld has many aspects that are helpful for this aspect of MusicXML. The exact syntax and organization of the S3W taxonomy is not so important from my point of view (and I have no particular issues with it either); the main concepts I'd like to see imported from S3W are the overall structure and the approach to substitution. I think it's a good starting point for thinking about what would work in MusicXML 3.0.

For me, the main question to be examined in SoundWorld is this:

Should sound IDs really encode things like staccato, marcato, etc? I find this problematic since it overloads the notion of an "instrument", and articulations are already covered elsewhere in MusicXML. I believe what the MusicXML spec is looking for is something more like InstrumentWorld ;) An instrument produces a family of timbres as dictated by performance directions -- it is not just one timbre. Of course, one may have an instrument that is played in a particular technique built around one specific articulation -- say, an entire flute part played in flutter-tongue. That's not problematic since it's in effect an exotic instrument sound, not an articulatory variant.

So what I am concerned about is a direction where specific articulations or performance techniques on individual notes wind up getting mapped to different sound IDs. That feels too granular to me and takes us away from the idea of an "instrument". It is likely to confuse implementers and lead to conflicting interpretations. Even the idea of ensemble vs. solo instruments is questionable -- couldn't these be variants, automatically selected by some programs to render performance directions identifying specific solo passages?

I did read the defense of this idea in the spec, but I didn't find myself buying into it at the end. It feels like a good way to organize sounds, but overly fine-grained for instruments. I suppose this is restating exactly what you pointed out -- SoundWorld is a sound taxonomy, not an instrument taxonomy. I guess I have some discomfort with that, while liking the rest of the concept quite a lot.

Best,

... . . . Joe

Joe Berkovitz President Noteflight LLC 84 Hamilton St, Cambridge, MA 02139 phone: +1 978 314 6271 www.noteflight.com
Hartmut Lemmel
 
Posts: 6
Joined: March, 2014
Reputation: 0

RE: Sibelius Standard SoundWorld (S3W) as a possible means of describing sounds in MusicXML 3.0

Postby Michael Good » Wed Feb 02, 2011 12:10 pm

Hi Daniel and Joe,

Thank you for the further discussion. I do think we are converging on the key design ideas that we need to go forward.

Daniel Spreadbury wrote:Likewise, it may be useful for some applications that consume MusicXML files to have optional information available about e.g. a text direction such as "sul pont."

We very much want to add this to MusicXML 3.0 as well. Most likely this will be through extensions to the sound element. I am envisioning that the top levels of the Sibelius / capella hierarchies will go with the score-instrument element, while the lower levels go with the sound element.

The information about solo vs. section is currently in the solo/ensemble elements within the score-instrument. In MusicXML 2.0, a switch between solo and tutti would involve a switch between instruments. So perhaps the solo and ensemble elements should be accessible in the sound element as well.

Now that the score-instrument design seems to be settling out, we should be able to move on with sound element design proposals pretty soon.

Best regards,

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

Re: Sibelius Standard SoundWorld (S3W) as a possible means of describing sounds in MusicXML 3.0

Postby Daniel Spreadbury » Wed Feb 02, 2011 2:38 pm

Michael Good wrote:The information about solo vs. section is currently in the solo/ensemble elements within the score-instrument. In MusicXML 2.0, a switch between solo and tutti would involve a switch between instruments. So perhaps the solo and ensemble elements should be accessible in the sound element as well.

Personally I think that the score-instrument element is the right place for the solo/ensemble distinction to be made. At least from a practical consideration, very few sounds in hardware or virtual playback devices have switches that go between solo and ensemble sounds without requiring another patch to be loaded, so when we were faced with this choice in Sibelius, we decided to treat solo vs. ensemble as an instrument-level distinction.

-- Daniel Spreadbury Senior Product Manager | Sibelius R & D

Avid Unit 20-23 City North Fonthill Road London N4 3HF United Kingdom

t +44 (0)20 7561 7915 | m +44 (0)7990 558997 | f +44 (0)20 7561 7888

Sibelius is Avid. Learn more at www.avid.com <http://www.avid.com/>

Avid Technology Europe Limited is a limited company registered in England and Wales. Registered No.: 2670844. Registered Office: Pinewood Studios, Pinewood Road, Iver Heath, Bucks SL0 0NH, England
Daniel Spreadbury, Steinberg
http://blog.steinberg.net
User avatar
Daniel Spreadbury
 
Posts: 13
Joined: March, 2014
Location: London, UK
Reputation: 0

Who is online

Users browsing this forum: No registered users and 2 guests

cron