Narrative: SongHelix: An online database tool for the discovery of classical song

 

I. Statement of need

There has never been a simple way to search through the vast number of art songs by poet, date of composition, thematic links, keywords, or utility. SongHelix brings this ever-evolving information available into one searchable location. It is an online tool that enables users to search through the vast repertoire of art songs in order to find just the right piece. The website’s immediate purpose is to give singers and teachers the primary online tool for finding related repertoire. A secondary, and broader purpose will be to help to revitalize the form of voice recital by allowing for creative programming through novel associations. The website makes it possible for the user to search song repertoire by terms customarily found in print reference, but SongHelix’s real power is the ability to search by “Feature” as well. When searching within the “Feature” category the user will be able to discover aspects of the songs such as: dreams, stages of life, moving water, Greek mythology, particular metaphors, etc.

We anticipate that the website’s primary users will be voice teachers at all levels (pre-college, undergraduate and graduate), voice students, collaborative pianists, and professional singers. Currently when teachers look for repertoire for their students, they usually rely on their own experience as a singer (though this method is poor since everyone’s voice is unique). If they are experienced teachers, they will rely on their knowledge of vocal repertoire and experience with a variety of singers. They might browse the contents of an anthology or recording, and they might ask colleagues or mentors for suggestions.

Teachers and singers might also turn to social media for crowdsourced opinions on repertoire selection and programming. A Facebook group called Professional Voice Teachers regularly sees requests from its members for suggestions of particular repertoire. Examples of requests from that group include:

  • A song less than 60 seconds for a young soprano, not in English
  • Songs about the life of Christ for a masters student
  • Duets for 2 mezzo-sopranos
  • Songs about hope for a soprano
  • Songs about love (but not limited to romantic love) for a recital in a museum

Our tool will also benefit song composers and DMA voice students and musicologists. Currently composers expend time and energy in becoming discovered or remaining relevant. With modern composers’ inclusion in the dataset, their music will be discovered alongside those of the great masters of the past. Additionally as the dataset grows, gaps in song scholarship become apparent. These gaps are ideal for catalyzing new research in the field for faculty or doctoral students’ lecture recitals or theses.

 

Previous work

Books that facilitate the discovery of song usually fall into two categories: guides to a particular repertoire and attempts to give a broad overview of all vocal repertoire. These mountains of scholarship are invaluable to singers, but they are static and not comprehensive. By virtue of their medium, they cannot keep pace with discovery or creation of song. Examples of excellent, but narrowly focused references are: A singer’s guide to the American art song, 1870-1980 by Victoria Etnier Villamil, A Shakespeare Music Catalogue in 5 volumes by Gooch and Thatcher; Rachmaninoff’s complete songs: a companion with texts and translations by Richard Sylvester; The Beethoven song companion by Paul Reid; and German poetry in song: an Index of Lieder by Lawrence Snyder. Broadly focused references include: From Studio to Stage by Barbara Doscher, The art of the song recital by Emmons and Sonntag; Singer’s Repertoire in 5 volumes by Berton Coffin; and Music for the Voice by Sergius Kagen.

There are several valuable contributions of online reference. Though discovery of repertoire is possible through many of them, this is not their focus. Lieder.net is a vast database of song texts and translations. Artsongcentral.com is a database of 250 songs that is “principally an archive and directory of free, printable sheet music for singers and voice teachers. An emphasis is placed on standard classical and traditional repertoire.” Composerdiversity.com is a new database that aids in the discovery of classical music by composers of underrepresented groups. The site with the most similarity to SongHelix is musicaltheatersongs.com. Here the user can search for appropriate repertoire from the genre of musical theater, though this site does not offer the granularity of keyword or the data visualization of SongHelix.

 

II. Project design

The following section details the work plan, goals, outcomes, assumptions, and risks of the project. The duration of this phase of our project is August 2018 to December 2019.

Project goals

  • Create a dataset that includes all recital-appropriate song repertoire
  • Develop a user interface that facilitates robust search and filter functions and an intuitive browsing experience through data visualization
  • Develop a network of contributing scholars and students to augment the dataset
  • Make the tool available to the public

Summary of work to date

  • Created database in Google Sheets. Accessible to data contributors. Backed up weekly through a third-party service. Dataset currently holds 3,500 songs.
  • Created advisory board of faculty, artists, and administrators. Members from  University of Utah, Indiana University, the University of Minnesota, and the Five Boroughs Music Festival.
  • Developed metadata procedures.
  • Created an online input method that sends information to a temporary spreadsheet for curation before inclusion into the dataset.
  • Developed guides to metadata procedures for off-site song scholars and students.
  • Created controlled vocabulary of keywords.
  • Developed network of data partnerships with interested song scholars and students.
  • Developed partnerships with Louise Toppin (University of Michigan), Christopher A. Reynolds (UC Davis), and Rob Deemer (SUNY Fredonia).
  • Incorporated links to Naxos Music Library audio files.
  • Hosting of the database and version 1.0 of the visualization on Amazon Web Services.

 

Products/outcomes

  • Grow the dataset to over 10,000 songs in the next 2 years.
  • Public release of SongHelix.
  • Present at regional and international conferences.
  • Publish 3 articles in peer-reviewed journals.

 

III. Broad impact

SongHelix will become the central method of song repertoire discovery. It will act as a hub from which the user will be directed to high quality recordings, online translations, websites that specialize in a deeper reading of the repertoire, print scholarship, and sheet music merchants. It will facilitate a galvanization of song enthusiasts and the world of classical voice.

With robust search and filtering characteristics and a novel data visualization component, SongHelix users will be able to make a highly targeted search or browse our nearly 6,000 keywords. For example, a person would be able to find all the songs by American composers about nature and joy, written between 1950 and 1960; or songs with texts by African American poets in a range that fits his or her voice within a specified level of difficulty. As the database grows, the number of links among songs will grow as well.

As important as a targeted search will be for our users, the value of browsing cannot be overstated. The data visualization component of SongHelix, in its first iteration, is represented as a node and tree system. It is important that the users be able to start from a single datapoint, and explore from there. For example, female composers to female composers/nature to female composers/nature/solitude. Additionally we have plans to graphically represent the search history so that a person knows how it is that they arrived at their results. Other ideas for interaction with the data include ways to bookmark songs and allowing users to suggest additions or corrections. Further, because the dataset will likely be overwhelming to the user, we are considering presenting a rotating curated subset of the data on the homepage.

 

Links to view

Version 1.0 of the search bar can be viewed here: http://bit.ly/SearchbarOne

Version 1.0 of the keywords visualization can be viewed here: http://bit.ly/visualizationOne