I am working on an iOS app that needs to recognize 4 different songs played on a music box. I tried using echoprint's fingerprint codegen (https://github.com/rexstjohn/echoprint-ios-sample), and while it works great if I play back an exact recording of the same song, it doesn't work if playing the music box "live". Evidently whatever differences in timing / volume / etc. arise between different instances of playing the music box lead it to detect as distinct songs.
Does anyone know a library or technique that would work best for this particular application? Or perhaps a way to hack echo print to be more forgiving for this music box application?
