Skip to content

AbstractVoice v0.10.0

Choose a tag to compare

View all tags
@github-actions github-actions released this 15 May 17:46
· 22 commits to main since this release
v0.10.0
a861ac4

Added

  • Added tts_engine="supertonic" for Supertone Supertonic 3 fixed-profile ONNX
    TTS, implemented through AbstractVoice's internal adapter/runtime rather than
    the external Supertonic Python SDK.
  • Added abstractvoice[supertonic] plus --supertonic prefetch/download
    commands, with Supertonic included in abstractvoice[apple],
    abstractvoice[gpu], abstractvoice[all-apple], and
    abstractvoice[all-gpu].
  • Exposed Supertonic built-in voice profiles (M1-M5, F1-F5) through the
    existing VoiceProfile abstraction and REPL/web profile surfaces.

Changed

  • Expanded local TTS language/catalog handling so Piper keeps its curated voice
    mapping while Supertonic exposes its 31-language text frontend.
  • CLI and web example auto TTS defaults are now install-aware: plain
    abstractvoice stays OpenAI/remote, while local/full installs that include
    Supertonic such as abstractvoice[all-apple] and
    abstractvoice[all-gpu] start on Supertonic.
  • TTS engine switches now go through VoiceManager.set_tts_engine(...), reset
    the active base voice/profile to that engine's default for the current
    language, and clear stale cloned-voice selections in the CLI/web examples.
  • REPL playback now warms the output stream in the background, and the audio
    player prefers the hardware default output sample rate before resampling
    synthesized audio, avoiding slow first local /speak startup on macOS audio
    devices that reject uncommon sample rates.
  • Updated dependency checks, REPL/web help, install docs, model-management docs,
    and licensing notices for the Supertonic optional model artifacts.

Removed

  • Removed the abstractvoice[local] install alias. Use abstractvoice[apple]
    or abstractvoice[gpu] for platform local stacks, abstractvoice[all-apple]
    or abstractvoice[all-gpu] for those stacks plus the web example, or compose
    granular extras such as abstractvoice[supertonic,stt,audio-io].

Fixed

  • Fixed the REPL base-TTS availability check so selecting Supertonic reports the
    Supertonic install/prefetch hint instead of a Piper voice-model error.
Assets 4
Loading