A Query by Humming system allows the user to find a song by humming part of the tune. No musical training is needed.The idea is simple: you hum into the microphone. The computer records the hum and extracts certain features corresponding to the melody and rhythm characteristics, then it compares the features to the features of the songs in the database. Finally it returns a rank list of the songs or song segments most similar to the humming.
The technique is based on our paper published on ACM SIGMOD03 from Yunyue Zhu, Dennis Shasha and Xiaojian Zhao. The basic idea is to apply time series approaches to match the pitch contour of the humming to the MIDI’s in the database. The technique turned out to be very successful for the 53 most popular beatles songs we collected.
The goal is to build a reliable and efficent large-scale system that collects thousands of songs and response in seconds. Currently we are trying to improve our system’s accuracy, efficiency and scalability. This will be done by improving the time series approaches and combining useful approaches from areas such as singal processing, pattern recognition, database indexing and linguistics.
Click here to try it out
Link doesn’t work anymore 🙁 Would have loved to try it out. If 75% of people are off-key, that means only 25% gets to have precise returns. 🙂