Finding Exact and Approximate Repeating Patterns to Build Database for Content-Based Music Information Retrieval in Carnatic Music
- G Sai Naresh
- Archana Raghuvamshi
- Dara Vikram
as index terms. For better storagIn music information retrieval MIR the task is to find the relevant musical files based on a query given in the form of index terms. Existing Music Retrieval System uses metadata like singer name, movie name, album name etc. as index terms. For better storage and retrieval, music must be represented in digital form. The problem of converting raw audio to symbolic representation has a great significance. It is possible to extract a sequence of symbols taken from an alphabet corresponding to attributes of the music such as the duration and the pitch of a note available from its MIDI representation. The pitch of the notes is more relevant for MIR purposes. The proposed system is aimed at retrieval of music files based on its content instead of metadata. This requires indexing data of music files in terms of frequently occurring patterns represented in terms of note sequences. The database of such a sophisticated Music Retrieval System contains music files represented as note sequences. Hence, it is essential to develop an algorithm to convert automatically midi file into corresponding note sequences.
G Sai Naresh, Archana Raghuvamshi, Dara Vikram. "Finding Exact and Approximate Repeating Patterns to Build Database for Content-Based Music Information Retrieval in Carnatic Music".INTERNATIONAL JOURNAL OF ENGINEERING DEVELOPMENT AND RESEARCH ISSN:2321-9939, Vol.4, Issue 3, pp.575-587, URL :https://rjwave.org/ijedr/papers/IJEDR1603095.pdf
Volume 4 Issue 3
Pages. 575-587