piperg2p documentation

• Installation

• Quick start

• Voice configuration

• Encoding

• eSpeak backends

• Pronunciation lexicons

• Core API

• Backend API

• Raw phonemes

• Phoneme types

• Compatibility policy

• Provenance

The project is an independent phoneme and ID frontend. It accepts prepared, speakable text and does not synthesize audio or own written-to-spoken semantic normalization.

Applications that need number, unit, currency, date, abbreviation, or other semantic expansion may prepare text with a separate package such as Spokenform before calling phonemize_prepared(). Spokenform is not a PiperG2P dependency.