Polyphonic monotimbral music transcription using dynamic networks
Pattern Recognition Letters - Special issue: Artificial neural networks in pattern recognition
Sound onset detection by applying psychoacoustic knowledge
ICASSP '99 Proceedings of the Acoustics, Speech, and Signal Processing, 1999. on 1999 IEEE International Conference - Volume 06
| Hi-index | 0.00 |
Custom ring tones for cellular phones have become extremely popular over the past several years, allowing users to customize their mobile computing experience, expressing their personality or enjoying their favorite tune. In the ring tone selection process, a user generally searches or browses for a ring tone by title in one of several ring tone databases and then purchases his selection. Research into query-by-humming systems may eventually enable users to select a song by singing a segment of it. This project, Sing-a-Ring, extends the user ring tone experience by allowing users to create their own ring tones by singing. Unlike current options to record a sound file and set that as a ring tone directly, this application uses an autocorrelation algorithm to convert the recorded sample to a MIDI file and then provides customization options, such as instrument and tempo selectors. This application is implemented on the JavaTM 2 Platform, Micro Edition (J2ME), and has been tested on a Motorola RAZR V3i phone. For the current implementation, due to format conversion issues, recordings must be created on a computer and then transferred to the phone. Nevertheless, the current application is a proof of concept with great potential, especially if access to AMR decoding libraries and the digital signal processing chip is available.