Singing Voice Transcription challenges
Big News: Our First In-Person Session is Here!
Singing Voice Transcription challenges
Presenter: Miguel Perez is an MIR researcher at Klangio. He mostly works on music transcription with emphasis on singing voice.
What to Expect: Automatic Singing Transcription (AST) is the process of converting a recorded vocal melody into digital musical notes. While singing is our most universal way of making music, it is incredibly difficult for computers to transcribe accurately. This talk explores how we translate the fluid, expressive nature of the human voice into a structured score, a task that remains a significant challenge in Music Information Retrieval. The presentation follows a three-part structure: first, an introduction to what music transcription is and why we need it for digital music services; second, a breakdown of the technical challenges, such as separating a voice from background instruments and handling different vocal textures; and finally, an overview of how machine learning is being used to solve these problems.
How to join:
This session will be held in person and will also be livestreamed: Sign-up to get the location
Future Sesssions:
If you are not a member, join Munich Music Labs to get full access to our knowledge-sharing sessions:
Subscribe to our Luma calendar to stay up to date with upcoming events!