Y3 | OrchView
OrchView
Y3 | Workgroup Summary
14 July 2021
13:30-15:00 EDT
Via Zoom
Workgroup Leaders:
F. Baril (OrchPlay))
Aims:
The main aims of this meeting were to summarize OrchView’s latest developments and discuss future developments. These included:
· the new features that were implemented based on the user requests following the Grouping Effects testing phase,
· the update of the Annotation XML data format to build ORCHARD database entries directly from an OrchView file,
· changes to the Metadata interface to enter information on the piece, score, and recording,
· Core engine / framework update and memory management efficiency improvements.
Discussion Points:
1. Summary of OrchView’s latest developments
a. Implementations of user requests following the Orchestral Grouping Effects testing phase
2. Demonstration of new OrchView features
a. The Annotation List may be sorted by annotator name, measure numbers, annotation type, date, etc.
b. Annotations automatically adjust themselves according to score layout modifications (ex: when hiding instrument or combining staves).
c. Direct link to taxonomy definitions by clicking on question marks
d. Improved score importing and staff detection to make process much faster
e. Works with drum timbre/drum sets. Orchestration all on one staff. Is there a way to do all this in OrchView?
i. You can put multiple instruments on the same staff. Not a lot of practice with this technique yet.
f. Export tool where you can export a movie of the annotations with music would be great. Or CSV?
i. Movie file yes, CSV not as useful. New programmer would be helpful in designing this.
g. The iPad version is not necessary for now—we should prioritize the desktop version before working on this.
3. Discussion on the implementation of Orchestration Technique Tools, Aims and Modes of Playing (Year 4)
a. The playing techniques are already integrated in OrchView for all orchestral families.
i. Aiming to have a version ready in January 2022
ii. A group of annotators will test 8 pieces
b. Orchestration functions will be tested in the same manner with a group of beta-testers
c. A new branch of orchestration techniques has been added: Timbral manipulations
iii. Timbral Modulations and Morphology
iv. Will have a separate tool with drop-down options
v. First draft of definitions for new terms is complete
vi. Bright-Dark and Dark-Bright timbral morphing techniques using examples from La mer
d. Aim: Finishing and implementing the new tools in OrchView (doubling/coupling tool, timbre morphology)
4. Discussion on MusicXML data-mining functions
a. The hope is to automatize many of the music XML functions
b. Rhythm detection in music XML is a challenge
c. Statistical analysis of the instrument combinations in music XML files.
i. a graphical representation of this would be interesting
d. Datamining instrument combinations to predict orchestral effects (blends, etc.)?
i. This requires a model for what makes things blend—this is so far unsuccessful. We can detect factors that contribute to blend (onset synchrony, parallel motion) but not the blend itself.
e. A tool to detect onset synchrony—same tool could address rhythm and blend?
i. We need to develop a way to handle rhythm first. Layer detection
ii. What textures are in the foreground and background?
f. Automating the doubling and intervals would be very useful!
g. Automatizing playing techniques
h. Datamining mood markings in addition to tempo markings?
i. Some information will be important to specific projects: what instruments play in a tessitura in certain rhythms across a full piece?
j. Mining for Klangfarbenmelodie—knowing what order the instruments sound in would help recognize patterns through the piece.
i. Large-scale trends—e.g., what percentage are these instruments playing together? Statistics on the time instruments play together, what combinations are favoured?
ii. Having more XML files would make this a lot better.
iii. Ability to compare timbral segments, search for recurrence of certain parameter pairings would be great.
iv. Formal analysis as soon as possible.
v. A tool that searches for motives, with changeable degrees of similarity. Ability to find exact repetitions, modified by pitch height, contour only, etc.
§ M. Zeller: ability to search a subsection of a motive would also be good.
k. Datamining the annotation data
l. Cross-taxonomic linking on the correspondence of techniques/effects etc.
5. Discussion about methods (interfaces and taxonomic links) to find correspondences or deviations between analyses from different taxonomies and on how this data could be presented to researchers
a. Cross-taxonomic linking of the correspondence of techniques/effects is dependent on datamining the annotation data.
b. A tree-like visualization is possible (ex: starting from one grouping effect and branching out to the techniques that gave rise to that effect)
6. Discussion on the future of OrchView - Features requests and suggestions
a. Custom tools?
i. will need to check with the new programmer. Would be a flexible custom tool—could make a palette.
ii. what types of tools will people make? That would be interesting
b. What is the timeline for the automatic detection of doublings?
i. The priority is for the orchestration techniques. Next ACTOR meeting will have the beginnings of that.
ii. One group wants to use Orchview to annotate doublings, but with custom tools
§ the doubling/coupling annotation tool is in development [demonstration of tool in development]
c. Will Orchview be available for Windows?
i. Our new programmer is a Windows guy and will likely make a version that will be available. Timeframe TBD
d. How is it available to students?
i. Version 1 is available to share and will always be free. It’s on the repository.
e. We need some tutorials on using OrchView V1
i. We need people to collaborate with us to make the tutorial videos.
f. People are interested in a timeline view of the annotations
i. maybe vertical format would be more manageable
g. isolating the different piano lines is challenging—too compact.
i. Note isolation with colour coding could be a solution.
Action Items:
1. Implementation of current designs for Orchestration Technique tools by January 2022
2. Primary internal testing phase starting in February 2022
3. Secondary testing phase with 8 annotators starting around March 2022. The scores annotated will be the same scores which were used for the Orchestral Grouping Effects testing phase.
4. In early 2022 once Orchestration Techniques have been implemented:
a. Rhythm detection development (necessary to find doublings and pairings)
b. Implementation of Modes of Playing dictionaries
5. Make a French version of the OrchView presentation.
Follow-Up:
1. The new programmer will likely make a version of OrchView for Windows that will be available. Timeframe TBD.
2. Some members are interested in alpha-testing new tool for automatic detection of doublings/couplings.
3. There is interest in datamining for Klangfarbenmelodie—knowing what order the instruments sound in would help recognize patterns through the piece.