Struct Chunker 

pub struct Chunker { /* private fields */ }

Audio chunker for segmenting streams into processing units.

Combines VAD boundaries with duration heuristics to produce processing-size chunks.

Implementations

impl Chunker

pub fn new(config: ChunkerConfig) -> Self

Create a new chunker with the given configuration.

pub fn default() -> Self

Create a chunker with default configuration (500ms chunks).

Alias for Chunker::new(ChunkerConfig::default()).

pub fn chunk( &self, audio: &[f32], sample_rate: u32, vad_segments: &[SpeechChunk], ) -> Result<Vec<ProcessedChunk>>

Segment audio into processing chunks aligned to VAD boundaries.

This variant assumes the VAD timestamps use the zero-based origin from AudioTimestamp::EPOCH. If the timestamps already include a stream offset, prefer Chunker::chunk_with_stream_start.

Arguments

• audio: Raw PCM samples (f32, normalized to [-1.0, 1.0])
• sample_rate: Audio sample rate in Hz (must be > 0)
• vad_segments: Speech boundaries from VAD analysis

Returns

Vector of ProcessedChunk covering the entire input duration with no gaps.

Errors

Returns Error::InvalidInput if:

• sample_rate is zero
• audio is empty
• VAD segments have invalid timestamps (end < start)

Performance

Target: <15ms total processing time per chunk generated.

pub fn chunk_with_stream_start( &self, audio: &[f32], sample_rate: u32, vad_segments: &[SpeechChunk], stream_start_time: AudioTimestamp, ) -> Result<Vec<ProcessedChunk>>

Segment audio into processing chunks with an explicit stream start time.

Use this variant when the VAD timestamps should be interpreted relative to a known stream start rather than zero-based.

use speech_prep::{Chunker, ChunkerConfig, SpeechChunk};
use speech_prep::time::{AudioDuration, AudioTimestamp};

let chunker = Chunker::new(ChunkerConfig::streaming());
let stream_start = AudioTimestamp::EPOCH;

// VAD emits wall-clock timestamps relative to the live stream
let segments = vec![SpeechChunk {
    start_time:  stream_start,
    end_time:    stream_start.add_duration(AudioDuration::from_millis(240)),
    confidence:  0.92,
    avg_energy:  0.4,
    frame_count: 48,
}];

let audio = vec![0.0; 3840]; // 240ms @ 16kHz
let chunks = chunker.chunk_with_stream_start(&audio, 16_000, &segments, stream_start)?;
assert_eq!(chunks.len(), 1);

Trait Implementations

impl Clone for Chunker

fn clone(&self) -> Chunker

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Chunker

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Copy for Chunker