Paper and Data

• Paper (arXiv)
• Data (Zenodo)
• Data (GitHub mirror)

Introduction

Welcome to the companion site for the paper “Reconstructing the Charlie Parker Omnibook using an audio-to-score automatic transcription pipeline”. This was work carried out by Xavier Riley, a PhD candidate on the AIM programme at QMUL.

This site contains links to Soundslice pages where you can preview all the data.

Here’s a video of it in action - remember, no human intervention was required for this transcription. As we discuss in the paper, there are issues with notes in the upper register and the rhythms are also over simplified here but we hope to address these problems in future work.

Live demo

You can try the pipeline out for yourself at Replicate. You will need to add payment details but you only pay for the time that the GPU is running. This currently works out to around $0.03 for a 2 minute transcription.

Model weights

Under preparation

Automatically Transcribed Scores

As described in the paper, it hasn’t been possible yet to transcribe the complete Omnibook using the pipeline yet. We provide a link to the 32 complete transcribed scores as MusicXML below:

Download Estimated MusicXML here

Synchronised Score Previews

• An Oscar For Treadwell
• Another Hairdo
• Anthropology
• Back Home Blues
• Barbados
• Billies’s Bounce
• Bird Gets The Worm
• Bloomdido
• Blue Bird
• Blues For Alice
• Buzzy
• Card Board
• Celerity
• Chasing The Bird
• Cheryl
• Confirmation
• Cosmic Rays
• Dewey Square
• Diverse
• Donna Lee
• KC Blues
• Kim 1
• Kim 2
• Au Privave No. 1
• Au Privave No. 2
• Chi Chi
• Ko Ko
• Laird Baird
• Marmaduke
• Mohawk (No. 1)
• Mohawk (No. 2)
• Moose the Mooche
• My Little Suede Shoes
• Now’s The Time (No. 1)
• Now’s the Time (No. 2)
• Ornithology
• Passport
• Perhaps
• Red Cross
• Relaxing With Lee
• Scrapple From the Apple
• Segment
• Shaw ‘Nuff
• Si Si
• Steeplechase
• The Bird
• Thriving on a Riff
• Visa
• Warming up a Riff
• Yardbird Suite

Feedback/Questions

We welcome feedback on the dataset - please direct this to Xavier Riley whose email address can be found in the footer of this page.

License

The dataset contains copyright restricted material and is shared with researchers under the following conditions:

• it may only be used by the individual signing below and by members of the research group or organisation of this individual. This permission is not transferable.
• it may be used only for non-commercial research purposes.

• it (or data enabling the its reproduction) may not be sold, leased, published or distributed to any third party without written permission from the it administrator.

• When research results obtained using it are publicly released (in the form of reports, publications, or derivative software), clear indication of the use of it shall be given, usually in the form of a citation of the following paper:

• X. Riley and S. Dixon (TBD), TBD.
• Queen Mary University of London shall not be held liable for any errors in the content of it nor damage arising from the use of it.
• The it administrator may update these conditions of use at any time.

Acknowledgements

The author is a research student at the UKRI Centre for Doctoral Training in Artificial Intelligence and Music, supported by UK Research and Innovation [grant number EP/S022694/1].