Research On The Technologies Of Music Retrieval Based On Content By Humming

Query by humming (QBH) is a particular case of Query by Content in multimedia databases. Query by humming system allows a user to find a song even if he merely knows the tune from part of the melody. The user simply hums the tune into a computer microphone, and the system searches through a database of songs for melodies containing the tune and returns a ranked list of search results. The user can then find the desired song by listening to the results.Most research into query by humming in the multimedia research community uses the notion of Contour information. Melodic contour is the sequence of relative differences in pitch between successive notes. It has been shown to be a method that the listeners use to determine similarities between melodies. Some systems have low retrieval precision because they rely on melodic contour information from the humming tune, which in turn relies on the error-prone note segmentation process. Some systems yield better precision when matching the melody directly from audio, but they are slow because of their extensive use of Dynamic Time Warping (DTW). Our research approach improves both the retrieval precision and speed compared to previous approaches.In order to avoid the note segmentation, we introduce the time series method. We treat music as a time series and exploit and improve well-developed techniques from time series databases to index the music for fast similarity queries. We describe the channel selection and skyline algorithms involved in the extraction of melodies from polyphonic midis. We detail the representation of tunes and hums as time series, the time warping distance metric used in the research literature to perform similarity comparisons between time series, and an efficient indexing method to prune the search space and return a ranked list of results. We improve on existing DTW technique by introducing a general framework for indexing time series. Through extensive experiments, we confirm our claims that the result is high scalability.