Nüsha OCR
How recognition, transliteration and translation work
Learn the differences between the three operations, how to review results and how examples are verified.
OCR extracts text from an image. Transliteration changes its writing system. Translation produces text in a target language. These operations may share a source but serve different purposes.
Updated:
How to use it
- Identify the source type and required operation.
- Review a representative sample.
- Save human corrections and export the format you need.
Evaluating a result
Compare source and output side by side. Check missing lines, mistaken letters, footnotes, numbers and reading order separately. A result on one document or profile cannot establish accuracy across all documents.
Published examples
Examples identify the source, usage rights, processing date, service level and human corrections. Raw and corrected output are presented separately. Unverified accuracy percentages and illustrative text are not presented as measured results.
Review your results
Generated text does not replace final verification. Uncertain readings and specialist passages require further review.