Struct WhisperStreamPcm 

pub struct WhisperStreamPcm { /* private fields */ }

Streaming PCM transcriber — direct port of stream-pcm.cpp main loop.

Two modes:

• Fixed-step (use_vad = false): process step_ms chunks with overlap.
• VAD-driven (use_vad = true): accumulate speech, transcribe on silence.

Implementations

impl WhisperStreamPcm

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

Create a new WhisperStreamPcm processor (simple VAD or no VAD).

pub fn with_vad( ctx: &WhisperContext, params: FullParams, config: WhisperStreamPcmConfig, reader: PcmReader, vad: WhisperVadProcessor, ) -> Result<Self>

Create a new WhisperStreamPcm processor with Silero VAD.

pub fn is_eof(&self) -> bool

Returns true when the underlying reader has hit EOF and all samples are drained.

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

Run one iteration of the main loop.

Returns Ok(Some(segments)) if transcription occurred, Ok(None) if waiting for more audio or sleeping, Err on fatal error.

Returns Ok(None) with no more audio when EOF + drained.

pub fn run<F>(&mut self, callback: F) -> Result<()>

where F: FnMut(&[Segment], i64, i64),

Run until EOF or error. Calls callback for each transcription.

pub fn total_samples(&self) -> i64

Get the total number of processed samples.

pub fn n_iter(&self) -> i32

Get the iteration count.