BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Playing By Ear: A Computational Approach - Seb Silas (Hanover Musi
 c Lab)
DTSTART:20250617T160000Z
DTEND:20250617T170000Z
UID:TALK231796@talks.cam.ac.uk
CONTACT:125293
DESCRIPTION:*Abstract*\n\nPlaying by ear is an essential skill in many mus
 ical styles\, supporting tasks like improvisation - and even sight-reading
 . Whilst jazz education often emphasises ear training\, more contemporaril
 y – and perhaps paradoxically - it also features visually-notated practi
 ce books to enable a more systematic approach to acquiring melodic represe
 ntations. However\, empirical research has yet to comprehensively explain 
 how playing by ear learning strategies (e.g.\, by sight vs. by ear) affect
  memory and recall performance. Additionally\, despite being a task apt fo
 r a computer\, a computerised way of managing the learning of large melodi
 c vocabularies by assessing musicians via produced musical input remains u
 nrealised. To address this\, we developed a software architecture to measu
 re and record playing by ear and singing skills in real time. Using large 
 corpora of jazz solos and sight-singing exercises\, we tested participants
  under different learning conditions (e.g.\, visual vs. auditory\, with or
  without singing first). With such data\, we developed scoring approaches 
 and sought complementary statistical modelling frameworks to analyse how p
 eople learn to play and sing melodies by ear.\n\nReflecting on these resul
 ts\, first\, I will discuss how various cognitive and musical factors—su
 ch as general working memory\, musical training\, and item-level melodic c
 haracteristics—shape the ability to play melodies by ear. Additionally\,
  I will examine learning patterns and how different learning strategies im
 pact the later recall of melodies. Second\, I will discuss how these resul
 ts can be incorporated into a computational model which predicts how well 
 somebody might be able to play a given melody by ear based on these factor
 s. Specifically\, I will frame my discussion in terms of item response the
 ory-inspired approaches and suggest a musical “DASH” (Difficulty\, Abi
 lity\, Study History\; Mozer & Lindsey\, 2017) model. This approach helps 
 manage melodic item banks more effectively\, making it easier to develop m
 elodic skills in both playing and singing by ear. Ultimately\, this resear
 ch can help design better educational tools which integrate aural learning
  with structured practice. In this light\, finally\, I will present two pr
 ototype computerised learning applications: Slonimsky — to improve playi
 ng by ear skills — and Songbird\, for melodic singing skills.\n\n*Biogra
 phy*\n\nSeb Silas is a PhD Researcher at Hanover Music Lab\, supervised by
  Reinhard Kopiez and Daniel Müllensiefen. He researches computational app
 roaches to musical learning and memory\, with a particular emphasis on lea
 rning melodies and playing by ear. He is an active music teacher and saxop
 honist\, improviser and composer\, most notably with the band Don’t Prob
 lem. He draws upon his experience in these fields in his academic research
 .\n\n*Zoom link*\n\nhttps://zoom.us/j/99433440421?pwd=ZWxCQXFZclRtbjNXa0s2
 K1Q2REVPZz09 (Meeting ID: 994 3344 0421\; Passcode: 714277)
LOCATION:CMS computer room\, Faculty of Music (11 West Road\, Cambridge\, 
 CB3 9DP)
END:VEVENT
END:VCALENDAR