Transcoda achieves state-of-the-art zero-shot OMR with an 18.46% OMR-NED error rate on synthetic scores and 63.97% on historical Polish scans using a 59M model trained in 6 hours via synthetic data, kern normalization, and grammar decoding.
An empirical evaluation of end-to-end polyphonic optical music recognition
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
dataset 1
citation-polarity summary
fields
cs.CV 2years
2026 2verdicts
UNVERDICTED 2roles
dataset 1polarities
use dataset 1representative citing papers
A CNN using ResNet-v2-style residual bottleneck blocks and multi-scale dilated convolutions followed by BiGRU and CTC loss achieves SeER of 7.52% and SyER of 0.45% on the Camera-PrIMuS dataset for optical music recognition.
citing papers explorer
-
Transcoda: End-to-End Zero-Shot Optical Music Recognition via Data-Centric Synthetic Training
Transcoda achieves state-of-the-art zero-shot OMR with an 18.46% OMR-NED error rate on synthetic scores and 63.97% on historical Polish scans using a 59M model trained in 6 hours via synthetic data, kern normalization, and grammar decoding.
-
A High-Accuracy Optical Music Recognition Method Based on Bottleneck Residual Convolutions
A CNN using ResNet-v2-style residual bottleneck blocks and multi-scale dilated convolutions followed by BiGRU and CTC loss achieves SeER of 7.52% and SyER of 0.45% on the Camera-PrIMuS dataset for optical music recognition.