vrr/agent-zero
2
0
Fork 0
mirror of https://github.com/agent0ai/agent-zero.git synced 2026-10-14 03:38:53 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 3
agent-zero/plugins/_kokoro_tts/README.md
Alessandro ce48667e98
Some checks failed
Build And Publish Docker Images / plan (push) Has been cancelled
Details
Build And Publish Docker Images / build (push) Has been cancelled
Details
Add Paradee as the default voice in Local TTS 
Offer Paradee Heart alongside Kokoro voices in the existing bundled TTS
plugin. Default unconfigured settings to Paradee while preserving saved
Kokoro voices, blends, speed and activation choices. Switching engines
replaces the voice selection without a separate plugin or startup dialog.

Add lazy CPU ONNX synthesis with cached, pinned model assets and split
long input without dropping phonemes. Make status and preload follow the
selected engine, and present both engines through the Local TTS UI.

Prepare Paradee dependencies in hooks.py on plugin install/update,
enabled-plugin startup and re-enable. Reuse installed packages, preserve
browser fallback on setup failure and keep requirements.txt unchanged.
Update plugin documentation and regression coverage.

Validation: 44 focused tests passed in both host and Docker framework
runtimes. Live synthesis and voice-picker switching passed on localhost:32081.
git diff --check passed. Fresh-image builds and actual missing-package
installation were not tested.
2026-10-08 21:36:02 +02:00

1.5 KiB
Raw Permalink Blame History

Local TTS (Kokoro & Paradee)

Built-in local speech synthesis, retaining the _kokoro_tts plugin identity.

Behavior

  • Defaults to Paradee-8M v1.0, a small CPU model with one American English Heart voice.
  • Select Paradee · Heart in the voice picker or choose a Kokoro voice. Paradee cannot be blended; selecting it replaces the current Kokoro selection.
  • Kokoro supports single voices, comma-separated equal blends, and optional in-memory weighted blends.
  • Existing saved voices, blends, speed, and enable/disable preferences are preserved. No startup choice dialog is shown.
  • Keeps browser-native speechSynthesis as the fallback path when disabled or synthesis fails.
  • Prepares Paradee's ONNX Runtime and English phonemizer dependencies through hooks.py on installation, updates, startup, and re-enable. Already-installed dependencies are reused; speech synthesis never installs packages.
  • Uses Kokoro's existing voice-pack cache and does not create persistent blend files. Paradee downloads its pinned ONNX model and vocabulary on first use into usr/plugins/_kokoro_tts/paradee/; later use works offline.

Config

  • voice: paradee (default), a Kokoro voice identifier, or a comma-separated Kokoro blend
  • voice_weights: optional mapping of voice identifiers to positive weights; when present, it defines the active blend
  • speed: playback speed multiplier for either model (default 1.1)

Routes

  • POST /api/plugins/_kokoro_tts/synthesize
  • POST /api/plugins/_kokoro_tts/status