Skip to main content

FrameGenerator

ftts_core

Trait FrameGenerator 

Source 
pub trait FrameGenerator {
    // Required methods
    fn begin_utterance(
        &mut self,
        prepared: &PreparedText,
    ) -> Result<(), GenerationError>;
    fn next_frame(&mut self) -> Result<Option<CodeFrame>, GenerationError>;
}
Expand description

Model-specific autoregressive frame generation behind the blocking engine facade.

Like TextPreparer, ftts-core owns only the boundary: the model crate implements the prompt assembly, talker forward, and 15-step microdecoder behind these two calls, and the engine owns admission, budgets, cancellation, and observer events around them.

Required Methods§

Source

fn begin_utterance( &mut self, prepared: &PreparedText, ) -> Result<(), GenerationError>

Prepares per-utterance state (prompt assembly and talker prefill) for one request.

Source

fn next_frame(&mut self) -> Result<Option<CodeFrame>, GenerationError>

Produces the next 16-code frame, or None once the model emits codec EOS.

Dyn Compatibility§

This trait is dyn compatible.

In older versions of Rust, dyn compatibility was called "object safety".

Implementors§