wsola 0.1.0

Pitch-preserved audio time-stretch (WSOLA), pure Rust, no C
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
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218

//! Public integration tests: drive `wsola` through its public API exactly as a
//! consumer (pdcst) does, and assert the properties that matter -
//! duration scales with tempo, pitch does NOT, output stays finite and bounded,
//! and streaming equals one-shot.

use wsola::{Config, TimeStretch, stretch};

const SR: u32 = 44_100;

/// A mono sine of `freq` Hz for `secs` seconds at 44.1 kHz.
fn sine(freq: f32, secs: f32) -> Vec<f32> {
    let n = (SR as f32 * secs) as usize;
    (0..n)
        .map(|i| (2.0 * std::f32::consts::PI * freq * i as f32 / SR as f32).sin())
        .collect()
}

/// Estimate the dominant frequency of a mono signal from its zero-crossing rate.
/// For a clean sine this is accurate to a few percent, which is all we need to
/// tell "pitch preserved" from "pitch shifted".
fn dominant_freq(signal: &[f32]) -> f32 {
    // Skip the fade-in/out edges where overlap-add ramps amplitude.
    let a = signal.len() / 8;
    let b = signal.len() - signal.len() / 8;
    if b <= a + 2 {
        return 0.0;
    }
    let seg = &signal[a..b];
    let mut crossings = 0usize;
    for w in seg.windows(2) {
        if (w[0] <= 0.0 && w[1] > 0.0) || (w[0] >= 0.0 && w[1] < 0.0) {
            crossings += 1;
        }
    }
    // Two zero-crossings per cycle.
    let secs = seg.len() as f32 / SR as f32;
    crossings as f32 / 2.0 / secs
}

fn rms(signal: &[f32]) -> f32 {
    if signal.is_empty() {
        return 0.0;
    }
    (signal.iter().map(|s| s * s).sum::<f32>() / signal.len() as f32).sqrt()
}

#[test]
fn duration_scales_inversely_with_tempo() {
    let input = sine(220.0, 2.0);
    for &tempo in &[0.5f32, 0.75, 1.0, 1.5, 2.0] {
        let out = stretch(&input, SR, 1, tempo).unwrap();
        let expected = input.len() as f32 / tempo;
        let ratio = out.len() as f32 / expected;
        // Within a few percent (boundary frames aside).
        assert!(
            (0.9..=1.1).contains(&ratio),
            "tempo {tempo}: got {} samples, expected ~{expected} (ratio {ratio:.3})",
            out.len()
        );
    }
}

#[test]
fn pitch_is_preserved_across_tempo() {
    let freq = 440.0;
    let input = sine(freq, 2.0);
    let base = dominant_freq(&input);
    assert!(
        (base - freq).abs() / freq < 0.03,
        "sanity: input reads {base} Hz"
    );

    for &tempo in &[0.5f32, 0.75, 1.25, 1.5, 2.0] {
        let out = stretch(&input, SR, 1, tempo).unwrap();
        let f = dominant_freq(&out);
        // The whole point: frequency must NOT move with tempo.
        assert!(
            (f - freq).abs() / freq < 0.05,
            "tempo {tempo}: pitch shifted to {f} Hz (want ~{freq})"
        );
    }
}

#[test]
fn identity_at_tempo_one_preserves_signal() {
    let input = sine(330.0, 1.0);
    let out = stretch(&input, SR, 1, 1.0).unwrap();
    // Length within a frame or two.
    assert!((out.len() as isize - input.len() as isize).unsigned_abs() < 4000);
    // Energy preserved (overlap-add of a Hann COLA reconstructs unit gain).
    let ri = rms(&input);
    let ro = rms(&out);
    assert!(
        (ro / ri - 1.0).abs() < 0.1,
        "rms changed too much: in {ri:.3} out {ro:.3}"
    );
}

#[test]
fn output_is_finite_and_bounded() {
    // A messy mix, not a clean tone.
    let input: Vec<f32> = (0..44_100)
        .map(|i| {
            let t = i as f32 / SR as f32;
            0.5 * (2.0 * std::f32::consts::PI * 200.0 * t).sin()
                + 0.3 * (2.0 * std::f32::consts::PI * 517.0 * t).sin()
        })
        .collect();
    let peak = input.iter().fold(0.0f32, |m, s| m.max(s.abs()));
    for &tempo in &[0.5f32, 1.0, 1.7, 3.0] {
        let out = stretch(&input, SR, 1, tempo).unwrap();
        assert!(
            out.iter().all(|s| s.is_finite()),
            "tempo {tempo}: non-finite output"
        );
        let out_peak = out.iter().fold(0.0f32, |m, s| m.max(s.abs()));
        assert!(
            out_peak <= peak * 1.5 + 1e-3,
            "tempo {tempo}: output peak {out_peak} blew past input peak {peak}"
        );
    }
}

#[test]
fn streaming_equals_one_shot() {
    let input = sine(300.0, 1.5);
    let one_shot = stretch(&input, SR, 1, 1.5).unwrap();

    // Feed the same input in irregular chunks, pulling small amounts as we go.
    let mut ts = TimeStretch::new(SR, 1).unwrap();
    ts.set_tempo(1.5);
    let mut streamed = Vec::new();
    let chunks = [17usize, 1024, 4096, 99, 8192, 333];
    let mut pos = 0;
    let mut ci = 0;
    while pos < input.len() {
        let n = chunks[ci % chunks.len()].min(input.len() - pos);
        ts.push(&input[pos..pos + n]);
        pos += n;
        ci += 1;
        streamed.extend(ts.pull(500));
    }
    streamed.extend(ts.flush());

    assert_eq!(
        streamed.len(),
        one_shot.len(),
        "streamed {} vs one-shot {}",
        streamed.len(),
        one_shot.len()
    );
    for (i, (a, b)) in streamed.iter().zip(&one_shot).enumerate() {
        assert!(
            (a - b).abs() < 1e-4,
            "sample {i}: streamed {a} != one-shot {b}"
        );
    }
}

#[test]
fn stereo_keeps_interleaving_and_pitch() {
    // Interleaved stereo: left 440 Hz, right 660 Hz.
    let n = SR as usize * 2;
    let mut input = Vec::with_capacity(n * 2);
    for i in 0..n {
        let t = i as f32 / SR as f32;
        input.push((2.0 * std::f32::consts::PI * 440.0 * t).sin());
        input.push((2.0 * std::f32::consts::PI * 660.0 * t).sin());
    }
    let out = stretch(&input, SR, 2, 1.5).unwrap();
    assert_eq!(out.len() % 2, 0, "interleaving broken");

    let left: Vec<f32> = out.iter().step_by(2).copied().collect();
    let right: Vec<f32> = out.iter().skip(1).step_by(2).copied().collect();
    assert!((dominant_freq(&left) - 440.0).abs() / 440.0 < 0.06);
    assert!((dominant_freq(&right) - 660.0).abs() / 660.0 < 0.06);
}

#[test]
fn reset_clears_state_and_resumes_like_new() {
    let a = sine(300.0, 0.5);
    let b = sine(300.0, 0.5);

    // Feed some audio, reset, then process `b` - result must equal processing
    // `b` on a fresh stretcher (reset leaves no residue from `a`).
    let mut ts = TimeStretch::new(SR, 1).unwrap();
    ts.set_tempo(1.5);
    ts.push(&a);
    let _ = ts.pull(usize::MAX);
    ts.reset();
    ts.push(&b);
    let mut after_reset = ts.pull(usize::MAX);
    after_reset.extend(ts.flush());

    let fresh = stretch(&b, SR, 1, 1.5).unwrap();
    assert_eq!(after_reset.len(), fresh.len());
    for (x, y) in after_reset.iter().zip(&fresh) {
        assert!((x - y).abs() < 1e-4);
    }
    // Tempo survives the reset.
    assert_eq!(ts.tempo(), 1.5);
}

#[test]
fn tempo_change_mid_stream_is_smooth() {
    let input = sine(400.0, 2.0);
    let mut ts = TimeStretch::with_config(SR, 1, Config::default()).unwrap();
    let mut out = Vec::new();
    ts.push(&input);
    // First half at 1.0, then switch to 1.5.
    out.extend(ts.pull(input.len() / 3));
    ts.set_tempo(1.5);
    out.extend(ts.pull(usize::MAX));
    out.extend(ts.flush());
    assert!(out.iter().all(|s| s.is_finite()));
    // Pitch still 400 Hz despite the tempo change.
    assert!((dominant_freq(&out) - 400.0).abs() / 400.0 < 0.06);
}