^{Alan D. Talbot SMDL} SMDL - Ten Years Later

Alan D. Talbot
8/21/96

1. Introduction

1. Music is a perfect prototypical problem for HyTime -- this is not surprising given that HyTime came from the original work on this music standard.
2. ^{<- DTD ->} This is a textbook example of applying HyTime to a real problem: derive a music DTD (SMDL - ISO/IEC 10743) from HyTime.
3. Representing music will use most if not all of HyTime.
4. Music has time: HyTime supports a time axis and scheduling of multiple tracks.
5. Music has multiple hierarchies: HyTime supports this with fcs location addressing.
6. Music requires resources: HyTime supports this with description tables.
7. Music requires heterogeneous lists of objects: HyTime supports this with derivable elements and attributes.
8. Music is divisible into five basic aspects; logical, gestural, visual, analytical, and timbral: HyTime supports these with complex hyperlinks and rendition (projection), except for the timbral information which was deliberately omitted from SMDL.

^Music 2. Music

1. Overview
   1. ^Spawn The music standard project (ANSI X3V1.8M) spawned HyTime.
   2. Music proves the validity of HyTime and also pushes its envelope.
   3. Music is a very abstract and very difficult problem; it will make everything else seem easy.
   4. Music models many other problems, so although it is a very small market, it is worthy of much attention.
2. ^{<- Object ->} Heterogeneous Lists of Objects
3. Library-like Resources
   1. Pitches
   2. Key Signatures
4. Time
   1. Sequences (Tracks) of Events
   2. Multiple Synchronized Tracks
5. Multiple Overlapping Hierarchies
   1. Metering
   2. Beaming
   3. Phrasing
   4. Dynamics
6. Five Basic Aspects
   1. Logical -- the music
      1. Somewhat like drama, but not like most other forms of art. More like the design of a bridge.
   2. Gestural -- the performance
      1. Can contain a lot of additional information
   3. Timbral -- the sound
      1. Overlaps with gestural
   4. Visual -- the score
      1. Contains all of the logical, and a lot more
   5. Analytical -- not part of the music; meta information
      1. Editorial information

3. Designing a Representation

1. Who is going to use it?
   1. Music Publishers (content providers)
      1. ^{<- WWW ->} They want to re-host content to Web, CD-ROM, etc.
      2. ^{<- Intranet ->} They need complex intranet services like document management and executive authoring.
      3. They need to merge music, graphics, body text, and the results of database searches.
      4. They want electronic publishing.
   2. Composers (content authors)
      1. They need remote collaboration
      2. They need re-use of arbitrary chunks
   3. Educators
      1. They want interactive remote education
      2. They want automated education
   4. Researchers
      1. They need to do arbitrary computations on and analyses of musical data
      2. They need remote collaboration
   5. Software Vendors (tool manufacturers)
      1. Notation
      2. Scanning (OCR)
      3. Composition
      4. Analysis
      5. Sequencing
2. What is it for?
   1. Interchange
   2. Archiving
   3. Re-use
      1. Education
      2. Research
      3. ^{<- Intranet ->} Executive Authoring - Intranet
      4. Electronic Publishing - Internet
3. What should be represented, and what should not?
   1. Music is arbitrarily complex - more so than most other applications - so we must reduce the problem.
   2. Logical domain is represented in a normalized way.
   3. Timbre is not represented. It is too hard for now. It is like color but much harder.
   4. ^{Analytical Domain} Gestural, Visual, and Analytical domains are provided with wrappers, but a representation is not specified. The Analytical domain is a good candidate for another HyTime derivation.

4. SMDL

1. Overview
   1. It is a conforming HyTime derived architecture
   2. ^{ANSI X3V1.8M} It is based on ANSI X3V1.8M work
   3. It is an ISO standard in the final editing phase
2. Structure
   1. Four Basic Domains
      1. Logical
      2. Gestural
      3. Visual
      4. Analytical
   2. Resources
      1. Pitch Gamut
   3. Heterogeneous Lists
      1. Threads
   4. Time
      1. FCS Axis
      2. Cantus Event Sequence
      3. Note and Rest Events
   5. Multiple Overlapping Hierarchies
      1. Metric Stress Pattern
   6. ^{<- Addressing -> <- Hyperlink ->} Complex Addressing and Hyperlinks
      1. ^{<- Addressing -> FCS} Addressing of ranges of FCS
      2. Connections between domains
   7. Rendering
      1. Rendering is not specifically addressed, although a wrapper is provided for the Gestural domain. A HyTime based gestural domain architecture could be defined in the future using the rendering facilities of HyTime.

5. Conclusions

1. Music is hard to represent, but worth it
2. HyTime is good
3. SMDL is going to be good
4. SMDL is proving the efficacy of HyTime
5. SMDL will drive improvements to HyTime and SGML