Changelog

Generated by releaseledger. Manual edits may be overwritten. Update releaseledger entries and rebuild this file instead.

[v0.1.5] - 2026-09-23

Changed

  • Changed phonemize_many to accept multiline text inputs with transparent CLI fallback

Quality

  • Added Termux-specific library discovery tests

[v0.1.4] - 2026-09-19

Added

  • Added native phoneme trace API path for CLI-equivalent IPA output

  • Added phoneme_output_api and phoneme_parity fields on RuntimeInfo for backend diagnostics

  • Added native library discovery probes for phoneme trace symbols

  • Added backend parity checking development tool

Documentation

  • Documented native trace and native translation phoneme output paths

Quality

  • Improved phoneme parity tests for weak words, contractions, and phrase context

[v0.1.3] - 2026-09-18

Added

  • Added Mode type constraining backend selection to auto, native, or cli

  • Added native_probes property on EspeakRuntime to expose library selection diagnostics

Changed

  • Changed phonemize_many in native backend to set voice once and reuse it across all texts

  • Changed CLI clauses to use phonemize_many for batch processing instead of individual calls

Documentation

  • Documented Mode type and native_probes property in API reference

  • Fixed changelog formatting and generation

Quality

  • Improved CI workflows and release artifact checks

[v0.1.1] - 2026-09-17

Added

  • Added native voice selection with language-property fallback for locale-style selectors

  • Added shared whitespace normalization helper for backend output parity

  • Added batch input validation for phonemize_many

Changed

  • Changed parity tests with regional locale matrix, custom U+200D tie coverage, batch parity, and auto mode parity

Documentation

  • Documented native voice selector resolution algorithm (name-first, language-property-second)

  • Documented regional locale selector support and custom tie examples in usage guide

  • Documented auto mode compatibility for downstream consumers in migration guide

[v0.1.0] - 2026-09-16

Added

  • Added native and CLI eSpeak NG backends for IPA phonemization, batching, voices, clauses, and diagnostics

  • Added exact native clause metadata and grouped punctuation handling for best-effort clause processing

  • Added non-initializing eSpeak capability inspection with ordered native candidate probes

  • Added language-style voice resolution with deterministic standard and explicit MBROLA selection

Fixed

  • Fixed cross-platform native library discovery by normalizing near-executable paths

  • Fixed CLI fallback after native initialization failures and normalized native and CLI data paths

  • Fixed native runtime cleanup on garbage collection and exposed normalized version components

Documentation

  • Documented the Sphinx build and Python-version-compatible TOML packaging configuration

  • Documented package version, Python support, download, and coverage badges in the project README

  • Documented Termux installation and Android-specific handling for unavailable bundled libraries

  • Documented structured fallback diagnostics, exact clauses, and Piper consumer ownership boundaries

  • Documented voice request resolution and runtime lifecycle behavior