TTS plugin development

storage/plugins/tts/* · LGPL-3.0 · host apiVersion per manifest

Overview: ASR / TTS plugins.

Examples:

• tts-plugin-echo
• tts-plugin-fishspeech

Layout

storage/plugins/tts/<dir>/
  plugin.json
  index.js

Role

Synthesize spoken-script segments to audio the host can stitch. Surface voices/config via configSchema for Settings UI.

API prefix

/api/tts-plugins (list / rescan / enable / config) — same kit as Source.

Full Chinese reference

• /development/tts-plugin
• docs/development/tts-plugin.md

Host synthesis behavior

• Split on 。！？!? and newlines: one sentence → one synthesizeChunk call
• maxCharsPerRequest only hard-splits an oversized single sentence
• Style/mood tags are planned per sentence (global base + scene controls like laugh/serious/breath), not a single opening-only tag
• Host synthesizes segments sequentially, inserts adaptive inter-sentence pauses (~0.45–1.2s by length/punctuation/paragraph), merges audio, and writes podcast.srt