Introduction¶

we are interested in an interdisciplinary approach to singing voice; we are looking for a balance between deep learning/engineering techniques vs real-life questions, cognition, physiology, musicology, etc., between bottom-up and top-down approaches. We want to learn about everyone’s research and ideas, broaden our scope of singing voice related topics and inspire each other to new research ideas.

Quick introduction round of everyone
Dan: Was working on singing voice modelling and is interested in singing voice timbre, today mostly focused on birdsong and pitch detection
Daniel: Working on singing voice separation, detection, and lyrics alignment
Yukun: Interested in pitch detection, transcription tasks
Chris: Software developer, worked on Sonic visualiser and Tony pitch tracking, interested in melody and performance of melody
Courtney: Singing voice as psychological phenomenon, singing voice techniques
Brendon: Interested in voice accents
Emir: Phoneme recognition and alignment, and finding useful features to describe singing voice
Polina: Singer, ensemble leader, ethnomusicologist, broad singing voice research, interested in ethnic aspects of singing and aspects of singing voice that are not in music notation generally
Wenming: Singing voice detection models with shallow and deep learning, interested in the Machine Learning aspects

Discussion of group size and meeting times - every two weeks appears suitable, group size okay

Possible agenda points
Discuss papers
Invite speakers
Datasets, annotations - next theme meeting?
Present our own research to each other
Going for drinks
Reporting on conferences you visited
Propose topics (e.g. vocal production, source separation)
Bring questions in advance?
Possibly share a bibliography of papers?

Idea: everyone presents their research for about 10min. But only 2 people per meeting