Struct AudioAnalyzer 

pub struct AudioAnalyzer {
    fft: FFTAnalyzer,
    bass_att: f32,
    mid_att: f32,
    treb_att: f32,
    attenuation: f32,
    agc_bass_avg: f32,
    agc_mid_avg: f32,
    agc_treb_avg: f32,
}

Audio analyzer for extracting per-band volumes.

Fields

fft: FFTAnalyzer

bass_att: f32

Long-term smoothed bass — the bass_att variable preset equations read.

mid_att: f32

Long-term smoothed mid — the mid_att variable preset equations read.

treb_att: f32

Long-term smoothed treble — the treb_att variable preset equations read.

attenuation: f32

Exponential-smoothing retention factor for the _att channels. Closer to 1.0 = longer history. MD2’s _att channels are heavily smoothed so they represent recent average loudness, not instantaneous.

agc_bass_avg: f32

Per-band running average used by the AGC to steer the gain. Updated each frame from the raw pre-AGC band level; the AGC divides the target level by this to compute the gain factor. Floor at 0.05 so the gain doesn’t blow up on near-silent input.

agc_mid_avg: f32

agc_treb_avg: f32

Implementations

impl AudioAnalyzer

pub fn new(sample_rate: f32) -> Self

Create a new audio analyzer.

pub fn analyze(&mut self, samples: &[f32]) -> AudioLevels

Analyze audio samples and return per-band volumes.

Pipeline: FFT → raw band integral → AGC. The AGC keeps each band hovering near AGC_TARGET_LEVEL = 1.0 on typical content, so bass > 1.5 on the output unambiguously means “louder than usual” — the convention every MD2 preset’s beat-detection block is authored against. Without the AGC, quiet tracks float around 0.2-0.4 and never trigger BeatDetectionMode::HardCut1; loud tracks pin at the 8.0 clamp and beat detection saturates.

pub fn set_attenuation(&mut self, attenuation: f32)

Set attenuation factor.

pub fn reset(&mut self)

Reset attenuated values.

Trait Implementations

impl Default for AudioAnalyzer

fn default() -> Self

Returns the “default value” for a type.