Resources

Datasets, models, and algorithms developed by our research group.

Datasets

Annotated-VocalSet: A Singing Voice Dataset

Link to the dataset: https://doi.org/10.5281/zenodo.7061507

This dataset provides annotations for the VocalSet dataset, available at https://doi.org/10.5281/zenodo.1442513 .

The annotations include fundamental frequency contour, note onset, note offset, transitions, note F0, note duration, MIDI pitch, and lyrics.

VocalSet consists of more than 10 hours of monophonic recorded audio of professional singers in a variety of vocal techniques (n = 17) and several singers (m = 20) with 3560 WAV files. Although categories such as techniques, singers, tempo, and loudness are considered, sung notes were not annotated. This dataset aims to annotate VocalSet to make it more powerful for researchers.

Details: Faghih, Behnam, and Joseph Timoney. 2022. "Annotated-VocalSet: A Singing Voice Dataset" Applied Sciences 12(18): 9257. https://doi.org/10.3390/app12189257

Modelling notes’ pitch and duration in trained professional singers

Link to the model repository: GitHub Repository

This study models the range of a note’s duration and pitch according to its position in a piece of music by analysing several parameters in trained-professional singers’ behaviours.

The repository includes R codes and the dataset used for the paper: "Modelling note’s pitch and duration in trained professional singers". Open-access at https://doi.org/10.1186/s13636-024-00380-4 . The paper includes details of the code and dataset.

SmartMedian

Link to the algorithm code: GitHub Repository

Smart-Median is an algorithm for smoothing pitch contours, altering outliers in singing signals.

Details: Faghih, B.; Timoney, J. Smart-Median: A New Real-Time Algorithm for Smoothing Singing Pitch Contours. Appl. Sci. 2022, 12, 7026. https://doi.org/10.3390/app12147026
Onset Detection

Link to the algorithm code: GitHub Repository

This algorithm introduces a new method for detecting onsets, offsets, and transitions of notes in real-time solo singing performances. It identifies onsets and offsets by finding transitions between notes via trajectory changes in fundamental frequencies.

Details: Faghih, B.; Chakraborty, S.; Yaseen, A.; Timoney, J. A New Method for Detecting Onset and Offset for Singing in Real-Time and Offline Environments. Appl. Sci. 2022, 12, 7391. https://doi.org/10.3390/app12157391