NLP app idea: construct random songs by scraping lyrics websites and stringing together common phrases. It’s a Pandora night for me and here were a couple lyrics that struck me as particularly meaningful. Both by Regina Spektor, introduced to me by Pandora before she became (semi-)famous.
And then you take that love you made
And stick it into some
Someone else’s heart
Pumping someone else’s blood
- “On the Radio”
Beneath the stars came fallin’ on our heads
But they’re just old light, they’re just old light
- “Samson”
I love how she takes the beautiful image of stars falling on their heads and strips it bare of all romanticism and attached meanings, exposing them for what they are: old light.
Advertisement
In my master’s NLP class at UNC, one of the students’ projects was doing something similar. Using a rhyming dictionary, some rough syllable analysis, vaguely thematic results from the top N pages from google and a simple HMM he created almost-sensible lyrics that rhymed and were related to a query term. I found it really entertaining and amazingly impressive for a student project.
Though not NLP generated, Scott Adams at his Dilbert Blog asked his readers to create a song by each contributing one line. Then, after he posted the best version, some German band produced an MP3 of it here.
And I have no doubt that both systems (student project and Dilbertian crowdsourcing) led to far better songs than 99% of the stuff produced in any given year.