Struct WhisperStream 

pub struct WhisperStream { /* private fields */ }

Streaming transcriber — faithful port of stream.cpp main loop.

Two modes:

• Fixed-step (step_ms > 0): sliding window with overlap.
• VAD (step_ms <= 0): transcribe on speech activity.

Implementations

impl WhisperStream

pub fn new(ctx: &WhisperContext, params: FullParams) -> Result<Self>

Create with default config.

pub fn with_config( ctx: &WhisperContext, params: FullParams, config: WhisperStreamConfig, ) -> Result<Self>

Create with custom config.

pub fn feed_audio(&mut self, samples: &[f32])

Push samples into the internal buffer (replaces SDL capture).

pub fn process_step(&mut self) -> Result<Option<Vec<Segment>>>

Dispatch to fixed-step or VAD mode.

pub fn flush(&mut self) -> Result<Vec<Segment>>

Process all remaining audio in buffer.

pub fn reset(&mut self)

Clear buffers, counters, prompt tokens.

pub fn buffer_size(&self) -> usize

Samples currently in the internal buffer.

pub fn processed_samples(&self) -> i64

Total samples consumed from the buffer.