Both LegaliPy and SonoriPy are available in the JS NLP library Talisman .
The algorithm has been updated, yielding 99.9 % accuracy (1998/2000) on the first 1000 words of both Parzival and Iwein.
We are happy to report tremendous progress in developing a model for automatic scansion of MHG epic verse. The model currently has an accuracy of 89.4% and will be available for demo shortly.
The LegaliPy Demo is now live!
LegaliPy is SyllabiPy’s younger, more ambitious brother. It operates strictly on the Legality Principle, which states that only those onsets permitted word initially are permitted word internally. LegaliPy takes text of any language as input, extracts all the onsets present in the text, and then reads through that same text and syllabifies according to the legal onsets.
The advantage of LegaliPy is that it requires no knowledge of the language whatsoever. Although syllabifiers adapted to a specific language will always be more accurate, LegaliPy can be used to efficiently syllabify any language to a high degree of accuracy. This is great for non-linguists who are looking to perform analysis of a text without having to learn about syllabification.
SyllabiPy has been working hard on a new tool for automatic MHG scansion. The tool will be released in beta in the coming weeks.
Support for Classical Latin is in beta, for better accuracy macrons are necessary.
Rhyme scheme identification for true rhyme has been added.
SyllabiPy version 1.1 witnesses the addition of Old High German and Gothic, both in the beta stage. Please contact us if you should find any errors.