# orthography2ipa

> Cited, per-language JSON specs that map spelling to IPA, a language-agnostic engine that turns them into transcriptions and ranked pronunciation lattices, and distance metrics that compare languages on separate axes. No trained weights.

Start with AGENTS.md before you change anything. Every spec is a file in orthography2ipa/data/, and its fields are in orthography2ipa/data/SCHEMA.md.

## Entry points

- [README](README.md): what the library does, install, quick start
- [AGENTS.md](AGENTS.md): repository map, setup, hard rules, checklist before a push
- [Spec schema](orthography2ipa/data/SCHEMA.md): every JSON key the loader reads
- [Python API](docs/reference/api.md): every exported name
- [Command line](docs/reference/cli.md): every subcommand
- [Scripts](scripts/README.md), [benchmarks](benchmarks/README.md), [tests](tests/README.md): what each directory holds

## Documentation, in reading order

- [Getting Started](docs/guide/getting_started.md)
- [Lexicons and forced pronunciations](docs/guide/lexicon.md)
- [Distance Metrics](docs/guide/distance.md)
- [Known limitations](docs/guide/known_limitations.md)
- [Python API](docs/reference/api.md)
- [Command line](docs/reference/cli.md)
- [Language Registry](docs/reference/registry.md)
- [Public API stability](docs/reference/api_stability.md)
- [Architecture](docs/engine/architecture.md)
- [Tokenizer](docs/engine/tokenizer.md)
- [The pronunciation lattice](docs/engine/lattice.md)
- [Lattice PER](docs/engine/lattice_per.md)
- [Allophony](docs/engine/allophony.md)
- [Positional Graphemes](docs/engine/positional_graphemes.md)
- [Data Model](docs/specs/data_model.md)
- [Quality tiers](docs/specs/quality_tiers.md)
- [Ancestry System](docs/specs/ancestry.md)
- [IPA Reference](docs/specs/ipa_reference.md)
- [Bibliography](docs/specs/bibliography.md)
- [Writing a language spec](docs/contributing/spec_authoring.md)
- [Citation honesty](docs/contributing/citations.md)
- [Board regeneration](docs/contributing/board.md)
- [Engine changes](docs/contributing/engine.md)
- [Silver IPA correction](docs/contributing/silver_correction.md)
- [Benchmarks](docs/evaluation/benchmarks.md)
- [Benchmark gold datasets](docs/evaluation/datasets.md)
- [The pinned WikiPron mirror](docs/evaluation/wikipron_mirror.md)
- [TTS silver sets](docs/evaluation/silver_sets.md)
- [Gold defects](docs/evaluation/gold_defects.md)
- [Scoreboard](docs/evaluation/generated/scoreboard.md)
- [Comparison to other G2P systems](docs/evaluation/generated/comparison.md)
- [espeak-ng agreement](docs/evaluation/generated/espeak_agreement.md)
- [Lexicon-overlay scoreboard](docs/evaluation/generated/lexicon_scoreboard.md)
- [Spec diagnostics](docs/evaluation/generated/spec_diagnostics.md)
- [Language pages](docs/languages/index.md)

## Optional

- [Per-language pages](docs/languages/index.md): phonology notes for single languages and families
- [Changelog](CHANGELOG.md)
