# CTC Transcription Alignment of the Bullinger Letters: Automatic Improvement of Annotation Quality

arXiv: https://arxiv.org/abs/2508.07904 · HTML: https://mlsanity.com/paper/2508.07904/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 10 of its 27 references that are in the map.
- First posted on arXiv 11 Aug 2025.

## Builds on (most cited first, 10 of 10)

- [On the Properties of Neural Machine Translation: Encoder-Decoder Approaches](https://mlsanity.com/paper/1409.1259/) (2014)
- [3D-VisTA: Pre-trained Transformer for 3D Vision and Text Alignment](https://mlsanity.com/paper/2308.04352/) (2023)
- [TACo: Token-aware Cascade Contrastive Learning for Video-Text Alignment](https://mlsanity.com/paper/2108.09980/) (2021)
- [Visual Alignment Constraint for Continuous Sign Language Recognition](https://mlsanity.com/paper/2104.02330/) (2021)
- [See Finer, See More: Implicit Modality Alignment for Text-based Person Retrieval](https://mlsanity.com/paper/2208.08608/) (2022)
- [Unified Coarse-to-Fine Alignment for Video-Text Retrieval](https://mlsanity.com/paper/2309.10091/) (2023)
- [CVT-SLR: Contrastive Visual-Textual Transformation for Sign Language Recognition with Variational Alignment](https://mlsanity.com/paper/2303.05725/) (2023)
- [Best Practices for a Handwritten Text Recognition System](https://mlsanity.com/paper/2404.11339/) (2024)
- [Impact of Ground Truth Quality on Handwriting Recognition](https://mlsanity.com/paper/2312.09037/) (2023)
- [Nuremberg Letterbooks: A Multi-Transcriptional Dataset of Early 15th Century Manuscripts for Document Analysis](https://mlsanity.com/paper/2411.07138/) (2024)

## Built on by (most cited first, 1 of 1)

- [BullingerDB: A Dataset for Handwritten Text Recognition and Writer Retrieval](https://mlsanity.com/paper/2605.30235/) (2026)

## Abstract (arXiv)

Handwritten text recognition for historical documents remains challenging due to handwriting variability, degraded sources, and limited layout-aware annotations. In this work, we address annotation errors - particularly hyphenation issues - in the Bullinger correspondence, a large 16th-century letter collection. We introduce a self-training method based on a CTC alignment algorithm that matches full transcriptions to text line images using dynamic programming and model output probabilities trained with the CTC loss. Our approach improves performance (e.g., by 1.1 percentage points CER with PyLaia) and increases alignment accuracy. Interestingly, we find that weaker models yield more accurate alignments, enabling an iterative training strategy. We release a new manually corrected subset of 100 pages from the Bullinger dataset, along with our code and benchmarks. Our approach can be applied iteratively to further improve the CER as well as the alignment quality for text recognition pipelines. Code and data are available via https://github.com/andreas-fischer-unifr/nntp.