From IPA Source:
SongHelix has recently come to our attention, and we believe that it might be of interest to some of our readers. The following post is by Seth Keeton, Director of SongHelix.
As director of the recently released SongHelix, I would like to share a bit about the tool’s creation and some of its behind-the-scenes features.
I know I’m not alone among recital lovers to say that the most rewarding part of giving a recital is the intimacy and direct, unimpeded communication with the audience. If this is our goal I would argue that for artists, a recital is an opportunity to say something; the only way to say something is if you craft a recital with loving detail. This is the impulse behind SongHelix. In order to say something, you have to know what repertoire there is to choose from. The idea for SongHelix was born in a post-recital chat with one of my mentors, Adriana Zabala, during my doctoral studies at the University of Minnesota. Together we wished, “wouldn’t it be great if there was one place where all art songs lived and just the right one could be discovered with a quick search?!?” A lot of planning, discussions with partners, internal grants from the University of Utah, and 4 years of work has culminated in the tool at songhelix.com.
Between my original discussion with Adriana and the creation of the tool, I planned several recitals for myself and my students, always wishing this tool existed. Discovering repertoire for an outdoors recital with animals in attendance on songs about animals in English took several weeks. Finding songs on Church and State for a student of mine double-majoring in political science and music was really challenging. Less challenging, but still time-consuming was discovering songs about British landscapes for performance in a museum with a traveling exhibition of British landscape paintings. The online forums on facebook typically have one or two posts a day asking for repertoire suggestions. It was clear to me that this tool needed to exist.
I was advised early on to decide what metadata (information about the data, in this case about the song) to include from the beginning–capturing as much information as possible from the outset, because going back to add a piece of info for each record would be horrible. So we tried to think of all the things one would want to know about a song from the beginning. This includes 40 or so fields, especially Keywords and what I call Features.
Keywords are the important words that appear in the poem in English translation. Though there is a legitimate argument to be made about what words are important enough to be considered Keywords, we have erred on the side of including more words than fewer. You never know what a person will want to find. Keywords are also relatively objective, since they’re just the words that appear in the translation. Features is broadly, “what the song is about.” This category is subjective. It contains our interpretation of the poem’s themes, poetic devices like particular metaphors (seasons turning as aging, for example), utility (encore piece, wedding, etc.), and anything else we feel is an interesting tag. Having this space for subjective analysis helps us move past the question of the quality of the translation we’re using. If we indicate that the song is about Flowers as a Feature, then whether is the keywords are Blossoms or Blooms isn’t such a big problem. All the Keywords and Features go into a big list with three levels of categories and subcategories. These are represented as the nested “folders” of terms on the browsing side of the site.
This list of Keywords and Features is what librarians call a “controlled vocabulary.” Semi-jokingly, I often refer to our list as an “uncontrolled vocabulary.” Our list of 8,000+ Keywords and Features developed organically. If we choose a Keyword or Feature, we add it to the list. Unlike the Library of Congress or other ontologies, we didn’t begin with preferred and secondary terms. Through little bits of code, we continue to improve the site and have developed a neat way around this problem. When a user tries a keyword that we don’t use in the top search box, there is a quick behind-the-scenes check of wordnet (an online thesaurus service) that suggests a keyword we /do/use. Similarly we are working on an internal synonym checker so that a search for Forests will also show results with Woods, Copses, Groves, etc.
Another example of behind-the-scenes planning is that of the multiple spellings of Russian composers’ names. In this case, we do have a preferred term (the spelling from Grove’s) and secondary terms all the other spellings we could find around the web. When a secondary spelling of a Russian composer’s name is searched, the preferred spelling is suggested in the box.
Please stop by and check it out at www.songhelix.com. I know it’s a disappointing feeling if you don’t see a song you’re expecting to, but(!) that’s your opportunity to help grow the dataset. It’s simple to contribute songs via a tab at the top that leads you to our online submission forms. My team and I are working every day to make SongHelix better and better, but our goal to be a /comprehensive/ hub for song discovery won’t be possible without a lot of help from our song-loving friends like you.
Bass-baritone, Seth Keeton’s performances have been described by The New York Times as “driven” and “emotionally pointed.” He has performed roles on the stages of The Minnesota Opera, Glimmerglass Opera, Ft. Worth Opera, Central City Opera, Arizona Opera, Indianapolis Opera, Austin Lyric Opera and Opera Omaha, Chautauqua Opera, Mill City Summer Opera, and Theater Bremen in Bremen, Germany.
As an oratorio singer, Keeton has appeared in concert as the bass soloist in Mozart’s Requiem, Verdi’s Requiem, Brahms’ Ein Deutsches Requiem, Bach’s Magnificat and St. Matthew and St. John Passions, Haydn’s Creation, and Beethoven’s 9th Symphony. In 2006, he was a national finalist in the Metropolitan Opera National Council Auditions and has received awards from the Sullivan Foundation and the Eleanor McCollum Competition. [more]