ebur128-stream 0.2.0

Streaming, zero-allocation EBU R128 loudness measurement in pure Rust.
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

//! Extended-coverage tests added for v0.1.2:
//!
//! - Calibration at the new 22.05 / 32 kHz rates added in this release.
//! - Long-programme stress (`#[ignore]` by default; run with
//!   `cargo test --release -- --ignored stress`).
//! - Snapshot-during-streaming test that proves mid-programme polling
//!   doesn't perturb the final report.

use ebur128_stream::{AnalyzerBuilder, Channel, Mode};

fn empirical_amp_for(target_lufs: f64, fs: u32, channels: &[Channel]) -> f32 {
    let probe_amp = 0.5_f32;
    let n = (fs as f32 * 5.0) as usize;
    let omega = 2.0 * std::f32::consts::PI * 1000.0 / fs as f32;
    let probe: Vec<f32> = (0..n)
        .map(|i| probe_amp * (omega * i as f32).sin())
        .collect();
    let interleaved: Vec<f32> = if channels.len() == 1 {
        probe
    } else {
        let mut out = Vec::with_capacity(probe.len() * channels.len());
        for v in &probe {
            for c in channels {
                out.push(if matches!(c, Channel::Lfe) { 0.0 } else { *v });
            }
        }
        out
    };
    let mut a = AnalyzerBuilder::new()
        .sample_rate(fs)
        .channels(channels)
        .modes(Mode::Integrated)
        .build()
        .unwrap();
    a.push_interleaved::<f32>(&interleaved).unwrap();
    let lufs = a.finalize().integrated_lufs().unwrap();
    let scale = 10f64.powf((target_lufs - lufs) / 20.0) as f32;
    probe_amp * scale
}

fn run_at(fs: u32, amp: f32, channels: &[Channel]) -> f64 {
    let n = (fs as f32 * 5.0) as usize;
    let omega = 2.0 * std::f32::consts::PI * 1000.0 / fs as f32;
    let mono: Vec<f32> = (0..n).map(|i| amp * (omega * i as f32).sin()).collect();
    let interleaved: Vec<f32> = if channels.len() == 1 {
        mono
    } else {
        let mut out = Vec::with_capacity(mono.len() * channels.len());
        for v in &mono {
            for _ in channels {
                out.push(*v);
            }
        }
        out
    };
    let mut a = AnalyzerBuilder::new()
        .sample_rate(fs)
        .channels(channels)
        .modes(Mode::Integrated)
        .build()
        .unwrap();
    a.push_interleaved::<f32>(&interleaved).unwrap();
    a.finalize().integrated_lufs().unwrap()
}

#[test]
fn calibration_at_22050_hz() {
    let layout = [Channel::Center];
    let amp = empirical_amp_for(-23.0, 22_050, &layout);
    let lufs = run_at(22_050, amp, &layout);
    assert!((lufs - (-23.0)).abs() <= 0.1, "I = {lufs}");
}

#[test]
fn calibration_at_32000_hz() {
    let layout = [Channel::Left, Channel::Right];
    let amp = empirical_amp_for(-23.0, 32_000, &layout);
    let lufs = run_at(32_000, amp, &layout);
    assert!((lufs - (-23.0)).abs() <= 0.1, "I = {lufs}");
}

#[test]
fn calibration_at_88200_hz() {
    let layout = [Channel::Left, Channel::Right];
    let amp = empirical_amp_for(-23.0, 88_200, &layout);
    let lufs = run_at(88_200, amp, &layout);
    assert!((lufs - (-23.0)).abs() <= 0.1, "I = {lufs}");
}

#[test]
fn calibration_at_96000_hz() {
    let layout = [Channel::Left, Channel::Right];
    let amp = empirical_amp_for(-23.0, 96_000, &layout);
    let lufs = run_at(96_000, amp, &layout);
    assert!((lufs - (-23.0)).abs() <= 0.1, "I = {lufs}");
}

#[test]
fn calibration_at_192000_hz() {
    let layout = [Channel::Left, Channel::Right];
    let amp = empirical_amp_for(-23.0, 192_000, &layout);
    let lufs = run_at(192_000, amp, &layout);
    assert!((lufs - (-23.0)).abs() <= 0.1, "I = {lufs}");
}

#[test]
fn snapshot_during_streaming_doesnt_perturb_report() {
    use core::time::Duration;
    let fs = 48_000_u32;
    let layout = [Channel::Left, Channel::Right];
    let amp = empirical_amp_for(-23.0, fs, &layout);
    let n = (fs as f32 * 30.0) as usize; // 30 s mono
    let omega = 2.0 * std::f32::consts::PI * 1000.0 / fs as f32;
    let mono: Vec<f32> = (0..n).map(|i| amp * (omega * i as f32).sin()).collect();
    let stereo: Vec<f32> = mono.iter().flat_map(|s| [*s, *s]).collect();

    let mut a = AnalyzerBuilder::new()
        .sample_rate(fs)
        .channels(&layout)
        .modes(Mode::All)
        .expected_duration(Duration::from_secs(30))
        .build()
        .unwrap();
    let mut b = AnalyzerBuilder::new()
        .sample_rate(fs)
        .channels(&layout)
        .modes(Mode::All)
        .expected_duration(Duration::from_secs(30))
        .build()
        .unwrap();

    // a: poll snapshot every 100 ms.
    for chunk in stereo.chunks(9_600 * 2) {
        a.push_interleaved::<f32>(chunk).unwrap();
        let _ = a.snapshot();
    }
    // b: don't poll.
    b.push_interleaved::<f32>(&stereo).unwrap();

    let r_a = a.finalize();
    let r_b = b.finalize();
    let diff = (r_a.integrated_lufs().unwrap() - r_b.integrated_lufs().unwrap()).abs();
    assert!(diff < 1e-9, "polling perturbs result by {diff}");
}

/// Long-programme stress: 1 hour of audio at 48 kHz, all modes
/// engaged. Asserts no overflow / drift / NaN escape.
///
/// Run with: `cargo test --release -- --ignored stress`
#[test]
#[ignore = "stress: ~1 minute of CPU time in --release mode"]
fn stress_one_hour_no_drift() {
    use core::time::Duration;
    let fs = 48_000_u32;
    let layout = [Channel::Left, Channel::Right];
    let mut a = AnalyzerBuilder::new()
        .sample_rate(fs)
        .channels(&layout)
        .modes(Mode::All)
        .expected_duration(Duration::from_secs(3_600))
        .build()
        .unwrap();

    // Push 1 hour in 100 ms chunks. We synthesize on the fly to avoid
    // allocating a full-hour buffer.
    let omega = 2.0 * std::f32::consts::PI * 1000.0 / fs as f32;
    let chunk_frames = 4_800usize;
    let mut buf: Vec<f32> = vec![0.0; chunk_frames * 2];
    for chunk_idx in 0..36_000 {
        // 36 000 chunks × 100 ms = 1 hour
        for f in 0..chunk_frames {
            let n = (chunk_idx * chunk_frames + f) as f32;
            let v = 0.05 * (omega * n).sin();
            buf[f * 2] = v;
            buf[f * 2 + 1] = v;
        }
        a.push_interleaved::<f32>(&buf).unwrap();
    }

    let r = a.finalize();
    assert!(r.integrated_lufs().unwrap().is_finite());
    assert!(r.true_peak_dbtp().unwrap().is_finite());
    assert!(r.loudness_range_lu().unwrap().is_finite());
    assert!((r.programme_duration_seconds() - 3_600.0).abs() < 1e-3);
}