orthography2ipa

What it is and what it does

orthography2ipa is a linguistically motivated library for converting written text to IPA phonetic notation and measuring how languages relate to each other. It ships with data specs for 676 languages and 73 clade nodes covering major language families, reconstructed proto-languages, dialects, and creoles. Rather than learning phoneme-to-sound mappings from weights, the library uses declarative per-language JSON specs that define graphemes, phonemes, allophones, positional rules, and genealogical metadata. A shared, language-agnostic engine applies these specs to produce transcriptions and confidence scores.

The library's core insight is that language similarity is multidimensional: two languages can sound alike but write differently, or vice versa. It measures each axis independently—phonological inventory overlap, how the same text sounds across languages, how the same sounds are spelled, script typology, ancestry, temporal span, and geographic distance. When spelling is ambiguous, the engine builds a ranked lattice of possible pronunciations with costs, allowing downstream systems to refine choices via custom rescoring rules. It also handles edge cases like logographic scripts (Chinese Hanzi has no grapheme map but a full phonology) and reconstructed languages (Proto-Indo-European declares its phonemes directly).

Use it for:

• Transcribe text to IPA for linguistic research, speech synthesis, or phonetic analysis across hundreds of language varieties.
• Measure phonological and orthographic distance between language pairs to study mutual intelligibility or dialect relationships.
• Build multilingual text-to-speech systems that need per-language grapheme-to-phoneme rules without training separate models.
• Analyze how spelling conventions diverge from pronunciation in a single language or across related languages.
• Refine ambiguous transcriptions using custom rescoring rules for domain-specific or specialized phonetic contexts.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Converts spelling to IPA phonetic transcription and measures language similarity across independent axes: phonology, reading, spelling, script, genealogy, time, and geography.

Yes, if you need IPA transcription or language similarity metrics for research or NLP. The library is actively maintained, has no known vulnerabilities, and covers a broad language base. The Alpha status and early release history mean the API and data specs may shift; verify transcription accuracy for your specific languages before production use. Low install friction and permissive licensing make it a low-risk addition.

Install

orthography2ipa on PyPI

pip install orthography2ipa

uv add orthography2ipa

poetry add orthography2ipa

Installing orthography2ipa

Before you install

Active maintenance with a recent release (29 days old). Low install friction: pure Python wheel with three runtime dependencies (numpy, langcodes, ovos-spec-tools). Supports current Python versions (3.10–3.13).

License in practice

Apache-2.0 permissive license allows commercial and private use with attribution and liability disclaimers. No copyleft obligations.

Quickstart

pip install orthography2ipa

import orthography2ipa
result = orthography2ipa.transcribe("olá mundo", "pt")
print(result)  # 'oˈla ˈmũdu'

Requires Python 3.10 or later.

Verify before relying

• Actual accuracy of IPA transcriptions across the 676 supported languages and dialects.
• Performance characteristics when processing large text volumes or many languages.
• Stability of the API and data format given the Alpha development status.

Package facts

Evidence: orthography2ipa-7.30.0-py3-none-any.whl

Keywords: ipa, phonetics, phonology, grapheme, allophone, linguistics, nlp

Tags

More Linguistic packages