lookas 1.8.0

Perception aligned terminal-based audio spectrum visualizer
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199

use crate::dsp::{a_weighting, ema_tc};
use crate::filterbank::Tri;

pub struct SpectrumAnalyzer {
    pub spec_pow_smooth: Vec<f32>,
    pub filters: Vec<Tri>,
    pub bars_y: Vec<f32>,
    pub bars_v: Vec<f32>,
    pub eq_ref: Vec<f32>,
    pub db_low: f32,
    pub db_high: f32,
    /// holds raw dB values (phase 1), then a sorted copy for percentile tracking (phase 2).
    sort_scratch: Vec<f32>,
    /// holds dB values temporarily (phase 1–2), then final normalised bar targets (phase 3).
    pub bars_target: Vec<f32>,
    /// holds lateral-flow intermediate values for the spring-damper step.
    flowed_scratch: Vec<f32>,
}

pub struct FlowSpringParams {
    pub flow_k: f32,
    pub spr_k: f32,
    pub spr_zeta: f32,
}

impl SpectrumAnalyzer {
    pub fn new(half_fft_size: usize) -> Self {
        Self {
            spec_pow_smooth: vec![0.0; half_fft_size],
            filters: Vec::new(),
            bars_y: Vec::new(),
            bars_v: Vec::new(),
            eq_ref: Vec::new(),
            db_low: -60.0,
            db_high: -20.0,
            sort_scratch: Vec::new(),
            bars_target: Vec::new(),
            flowed_scratch: Vec::new(),
        }
    }

    #[inline]
    pub fn resize(&mut self, num_bars: usize) {
        if self.bars_y.len() != num_bars {
            self.bars_y = vec![0.0; num_bars];
            self.bars_v = vec![0.0; num_bars];
            self.eq_ref = vec![1e-6; num_bars];
            self.bars_target = vec![0.0; num_bars];
            self.sort_scratch = vec![0.0; num_bars];
            self.flowed_scratch = vec![0.0; num_bars];
        }
    }

    pub fn update_spectrum(
        &mut self,
        spec_pow: &[f32],
        tau_spec: f32,
        dt_s: f32,
    ) {
        let len = self.spec_pow_smooth.len().min(spec_pow.len());
        for (i, &pow) in spec_pow.iter().enumerate().take(len) {
            let incoming = pow.max(1e-12);
            let prev = self.spec_pow_smooth[i];

            // constant attack if energy rose, snap immediately;
            // smooth release if energy fell, decay via EMA.
            self.spec_pow_smooth[i] = if incoming >= prev {
                incoming
            } else {
                ema_tc(prev, incoming, tau_spec, dt_s)
            };
        }
    }

    /// compute per-band dB values and map them to normalised bar targets.
    pub fn analyze_bands(&mut self, dt_s: f32, gate_open: bool) {
        let filters_len = self.filters.len();

        // --- Phase 1 --- //
        for (i, tri) in self.filters.iter().enumerate() {
            let mut acc = 0.0f32;
            for &(idx, wgt) in &tri.taps {
                if idx < self.spec_pow_smooth.len() {
                    acc += self.spec_pow_smooth[idx] * wgt;
                }
            }
            let amp = acc.sqrt();

            // Perceptually accurate frequency-sensitivity curve that peaks at
            // ~3–4 kHz and rolls off at both ends.
            let amp_weighted = amp * a_weighting(tri.center_hz);

            self.eq_ref[i] =
                ema_tc(self.eq_ref[i], amp_weighted, 6.0, dt_s)
                    .max(1e-9);
            let rel = amp_weighted / self.eq_ref[i];

            self.bars_target[i] = 20.0 * rel.max(1e-12).log10();
        }

        // --- Phase 2 ---
        self.sort_scratch[..filters_len]
            .copy_from_slice(&self.bars_target[..filters_len]);
        self.update_db_range(filters_len, dt_s);

        // --- Phase 3 ---
        let low = self.db_low - 3.0;
        let high = self.db_high + 6.0;
        let range_inv = 1.0 / (high - low).max(12.0);

        if gate_open {
            for i in 0..filters_len {
                let mut v = (self.bars_target[i] - low) * range_inv;
                v = v.clamp(0.0, 1.0).powf(0.85);
                self.bars_target[i] = 1.0 - (1.0 - v).powf(1.6);
            }
        } else {
            for t in &mut self.bars_target[..filters_len] {
                *t = 0.0;
            }
        }
    }

    fn update_db_range(&mut self, len: usize, dt_s: f32) {
        if len == 0 {
            return;
        }

        // sort_scratch already holds a copy of the raw dB values (set by analyze_bands).
        self.sort_scratch[..len].sort_by(|a, b| a.total_cmp(b));

        let idx_low = ((len - 1) as f32 * 0.10).round() as usize;
        let idx_high = ((len - 1) as f32 * 0.90).round() as usize;

        let q10 = self.sort_scratch[idx_low];
        let q90 = self.sort_scratch[idx_high];

        self.db_low = ema_tc(self.db_low, q10, 0.30, dt_s);
        self.db_high = ema_tc(self.db_high, q90, 0.50, dt_s);
    }

    /// Advance bar positions by one frame using lateral energy diffusion followed
    /// by a spring-damper integration.  Reads targets from `self.bars_target`
    /// (written by `analyze_bands`)
    pub fn apply_flow_and_spring(
        &mut self,
        params: &FlowSpringParams,
        dt_s: f32,
        gate_open: bool,
    ) {
        let n = self.bars_y.len();
        if n == 0 {
            return;
        }

        if !gate_open {
            let tau_silence = 0.22f32;
            let a = (-dt_s / tau_silence).exp();

            for i in 0..n {
                self.bars_y[i] *= a;
                self.bars_v[i] = 0.0;
                if self.bars_y[i] < 0.001 {
                    self.bars_y[i] = 0.0;
                }
            }
            return;
        }

        // lateral energy diffusion into pre-allocated scratch.
        for i in 0..n {
            let left = if i > 0 {
                self.bars_y[i - 1]
            } else {
                self.bars_y[i]
            };
            let right = if i + 1 < n {
                self.bars_y[i + 1]
            } else {
                self.bars_y[i]
            };
            let flow =
                params.flow_k * (left + right - 2.0 * self.bars_y[i]);
            self.flowed_scratch[i] =
                (self.bars_target[i] + flow).clamp(0.0, 1.0);
        }

        let c = 2.0 * params.spr_k.sqrt() * params.spr_zeta;

        for i in 0..n {
            let a = params.spr_k
                * (self.flowed_scratch[i] - self.bars_y[i])
                - c * self.bars_v[i];
            self.bars_v[i] += a * dt_s;
            self.bars_y[i] = (self.bars_y[i] + self.bars_v[i] * dt_s)
                .clamp(0.0, 1.0);
        }
    }
}