summaryrefslogtreecommitdiff
path: root/src/window_function.rs
blob: 3fbcedf3fc62dcdaaa99e6b31fd956fcfe92948c (plain)
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
use crate::window_size::WindowSize;
use std::{collections::VecDeque, f32::consts::PI, mem};

pub trait WindowFunction: Send {
    fn apply(&mut self, data: &mut VecDeque<f32>);
    fn reverse(&mut self, data: &mut [f32]);
    fn needed(&self) -> usize;
}

pub struct RectangularWindow {
    window_size: usize,
}

impl RectangularWindow {
    pub fn new(window_size: &WindowSize) -> Self {
        Self {
            window_size: window_size.inner(),
        }
    }
}

impl WindowFunction for RectangularWindow {
    fn apply(&mut self, _data: &mut VecDeque<f32>) {}

    fn reverse(&mut self, _data: &mut [f32]) {}

    fn needed(&self) -> usize {
        self.window_size
    }
}

pub struct HannWindow {
    window_size: usize,
    function: Vec<f32>,
    normalize: Vec<f32>,
    previous: Vec<f32>,
    previous_overlap: Vec<f32>,
}

impl HannWindow {
    pub fn new(window_size: &WindowSize) -> Self {
        let window_size = window_size.inner();
        let half_window_size = window_size / 2;
        let window_size_f32 = window_size as f32;

        // 50% overlap
        let previous_overlap = vec![0.0; half_window_size];

        // with capacity is important for the first needed samples
        let previous = Vec::with_capacity(window_size);

        let function = (0..window_size)
            .map(|n| {
                let n = n as f32;

                0.5 * (1.0 - f32::cos((2.0 * PI * n) / (window_size_f32 - 1.0)))
            })
            .collect::<Vec<f32>>();

        let normalize = (0..half_window_size)
            .map(|i| {
                (function[i] * function[i])
                    + (function[i + half_window_size] * function[i + half_window_size])
            })
            .collect::<Vec<f32>>();

        Self {
            window_size,
            function,
            normalize,
            previous,
            previous_overlap,
        }
    }
}

impl WindowFunction for HannWindow {
    fn apply(&mut self, data: &mut VecDeque<f32>) {
        let half_window_size = self.window_size / 2;

        // the full window size is needed to window over the samples the first time
        if self.previous.is_empty() {
            // store the latter half for when only half the size is needed
            for sample in data.iter().skip(half_window_size) {
                self.previous.push(*sample);
            }

            // apply the function
            for (i, sample) in data.iter_mut().enumerate() {
                *sample *= self.function[i];
            }

            return;
        }

        // In a real time "sliding window" it's necessary to add in the previous samples
        // to get the full resolution
        for sample in self.previous.iter().rev() {
            data.push_front(*sample);
        }

        self.previous.clear();

        // keep this iteration's samples for the next window
        for sample in data.iter().skip(half_window_size) {
            self.previous.push(*sample);
        }

        // apply the function
        for (i, sample) in data.iter_mut().enumerate() {
            *sample *= self.function[i];
        }
    }

    fn reverse(&mut self, data: &mut [f32]) {
        let half_window_size = self.window_size / 2;

        for (sample, function) in data.iter_mut().zip(self.function.iter()) {
            *sample *= *function;
        }

        // add back in the previous applied window function
        // this will reverse the 50% overlap
        for (i, (previous, sample)) in self
            .previous_overlap
            .iter()
            .zip(data.iter_mut())
            .enumerate()
        {
            *sample = (*sample + previous) / self.normalize[i];
        }

        self.previous_overlap.clear();

        // copy this window's tail for future overlap
        for sample in data.iter().skip(half_window_size) {
            self.previous_overlap.push(*sample);
        }
    }

    fn needed(&self) -> usize {
        if self.previous.is_empty() {
            self.window_size
        } else {
            self.window_size / 2
        }
    }
}