Turkish şarkı vocal dataset is a collection of recordings of compositions from the vocal form şarkı. The recordings

are selected from a musicBrainz collection of Turkish music

http://musicbrainz.org/collection/544f7aec-dba6-440c-943f-103cf344efbb

The collection has annotations of lyrics. Each lyrical phrase is aligned to its corresponding segment in the audio.

Download

This dataset can be downloaded here.

The dataset

Audio music content

version 1

It features 10 performances of different compositions. Five are sung by male and 5 by female singer. The recordings are selected so that there is only a single main vocalist (or the intensity of the back vocalst is relatively low compared to the main vocalist). Accompanying instruments are mainly string ensembles. No percussive instruments are present. Audio is in .wav format.

Please cite the following publication for if you use the dataset in your work:

Dzhambazov, G., Şentürk S., & Serra X. (2014). Automatic lyrics-to-audio alignment in classical Turkish music. 4th International Workshop on Folk Music Analysis

version 2

It is an modification with some added and some omitted recordings of Version 1 and features 12 performances of 11 different compositions. 8 sung by female and 4 by male.

Please cite the following publication for if you use the dataset in your work:

Dzhambazov, G., & Serra X. (2015). Modeling of Phoneme Durations for Alignment between Polyphonic Audio and Lyrics. Sound and Music Computing Conference 2015.

Lyrical phrases annotations

The audio is segmented into one-section chunks (a section is nakarat, meyan etc.)

Each audio segment is aligned to the lyrical phrases. A phrase corresponds roughly to a musical

bar and contains 1 or 2 words.

An annotation file is in .TextGrid format of Praat.