Jan Van BalenI’m attending #ISMIR2024 this week, the music IR/ML/AI conference—hope to be sharing some thoughts from here
Do say hi if you’re here as well!
Deezer Research
Ondřej Cífka"Lyrics Transcription for Humans: A Readability-Aware Benchmark", accepted to #ISMIR2024, is now online:
https://arxiv.org/abs/2408.06370
We evaluated more models (Whisper v3, OWSM v3.1, AudioShake v3) on our benchmark and included plots detailing what kinds of errors different models make on lyrics transcription.
https://arxiv.org/abs/2408.06370
We evaluated more models (Whisper v3, OWSM v3.1, AudioShake v3) on our benchmark and included plots detailing what kinds of errors different models make on lyrics transcription.
Lyrics Transcription for Humans: A Readability-Aware Benchmark
Writing down lyrics for human consumption involves not only accurately capturing word sequences, but also incorporating punctuation and formatting for clarity and to convey contextual information. This includes song structure, emotional emphasis, and contrast between lead and background vocals. While automatic lyrics transcription (ALT) systems have advanced beyond producing unstructured strings of words and are able to draw on wider context, ALT benchmarks have not kept pace and continue to focus exclusively on words. To address this gap, we introduce Jam-ALT, a comprehensive lyrics transcription benchmark. The benchmark features a complete revision of the JamendoLyrics dataset, in adherence to industry standards for lyrics transcription and formatting, along with evaluation metrics designed to capture and assess the lyric-specific nuances, laying the foundation for improving the readability of lyrics. We apply the benchmark to recent transcription systems and present additional error analysis, as well as an experimental comparison with a classical music dataset.
The deadline to submit your research to the NLP4MusA #ISMIR2024 companion workshop has been extended to July 12th: https://sites.google.com/view/nlp4musa-2024/home . Both academic and industry submissions at the crossroads of #NLP and #MIR are welcomed
Looking forward to presenting our lyrics transcription benchmark at #ISMIR2024 in San Francisco! Our paper grew from last year's LBD to a full paper with more results, which has now been accepted, so stay tuned!
https://audioshake.github.io/jam-alt/
https://audioshake.github.io/jam-alt/