Expand description
sayd’s engine: queue, chunking, text cleanup, config and the state machine that drives them.
This crate deliberately depends on neither ONNX nor an audio backend nor
D-Bus. Synthesis reaches it through the Synthesizer trait and audio
through AudioSink, so the whole engine can be driven in a unit test.
Modules§
- audio
- How the engine hands samples to the speakers.
- chunk
- Split text into synthesis units.
- cleanup
- Turn selected text into something worth hearing.
- config
- Persistent settings. TOML at
$XDG_CONFIG_HOME/sayd/config.toml. - engine
- The engine: one owner for the queue, the synthesizer and the sink.
- handle
- Runs the engine on its own thread and gives everyone else a handle to it.
- queue
- The utterance queue and the policy that governs what a new submission does to whatever is already speaking.
- synth
- The seam between the engine and actual speech synthesis.