vox_box 0.3.0

A tool to process speech audio in Rust
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

extern crate num;
extern crate sample;

use std::f64::consts::PI;
use std::iter::Iterator;
use std::cmp::Ordering::*;

use sample::{Sample, FloatSample, FromSample};

pub trait RMS<S> {
    fn rms(&self) -> S;
}

impl<S: Sample> RMS<S> for [S] {
    fn rms(&self) -> S {
        let sum = self.iter()
            .fold(S::equilibrium(), |acc, &item: &S| {
                acc.add_amp(item.mul_amp(item.to_float_sample()).to_signed_sample())
            });
        (sum.to_float_sample() / (self.len() as f64).to_sample::<S::Float>())
            .sample_sqrt().to_sample::<S>()
    }
}

pub trait Amplitude<S> {
    fn amplitude(self) -> S;
}

impl<S: Sample> Amplitude<S> for S {
    fn amplitude(self) -> S {
        if self < S::equilibrium() {
            self.mul_amp(S::Float::from_sample(-1.0))
        } else {
            self
        }
    }
}

pub trait MaxAmplitude<S> {
    fn max_amplitude(&self) -> S;
}

/// Returns the maximum peak amplitude in a given slice of samples
impl<S: Sample> MaxAmplitude<S> for [S] {
    fn max_amplitude(&self) -> S {
        assert!(self.len() > 0);
        if self.len() == 1 {
            self[0].amplitude()
        } else {
            self[1..].iter().fold(self[0].amplitude(), |acc, elem| {
                let amp = elem.amplitude();
                match amp.partial_cmp(&acc) {
                    Some(Greater) => amp,
                    _ => acc
                }
            })
        }
    }
}

pub trait Normalize<S> {
    fn normalize_with_max(&mut self, max: Option<S>);
    fn normalize(&mut self) {
        self.normalize_with_max(None);
    }
}

impl<S: Sample> Normalize<S> for [S] {
    fn normalize_with_max(&mut self, max: Option<S>) {
        let scale_factor: <S as Sample>::Float = <S as Sample>::identity() / 
            max.unwrap_or(self.max_amplitude()).to_float_sample();
        for elem in self.iter_mut() {
            *elem = elem.mul_amp(scale_factor);
        }
    }
}

/// Filter
///
/// Preemphasis should give a 6db/oct boost above a particular center frequency
/// Factor is center `frequency / sample_rate`
pub trait Filter {
    fn preemphasis(&mut self, factor: f64) -> &mut Self; 
}

impl<S: Sample + FromSample<f64>> Filter for [S] {
    fn preemphasis<'a>(&'a mut self, factor: f64) -> &'a mut [S] {
        let mut last = self[self.len()-1];
        let filter = 2.0 * PI * factor;
        for x in self.iter_mut().rev().skip(1) {
            *x = x.add_amp(last.mul_amp(filter.to_sample::<S::Float>()).to_signed_sample());
            last = *x;
        }
        self
    }
}

#[cfg(test)]
mod tests {
    extern crate sample;

    use super::*;
    use super::super::*;
    use super::super::periodic::*;

    use sample::conv::ToSampleSlice;
    use sample::window::Window;

    fn sine(len: usize) -> Vec<f64> {
        let rate = sample::signal::rate(len as f64).const_hz(1.0);
        rate.clone().sine().take(len).collect::<Vec<[f64; 1]>>().to_sample_slice().to_vec()
    }

    #[test]
    fn test_pe() {
        let mut sine = sine(32);
        sine.preemphasis(0.1f64); // preemphasize at 0.1 * sampling rate
    }

    #[test]
    fn test_window_autocorr() {
        let lag_window: Window<[f64; 1], HanningLag> = Window::new(16);
        let data: Vec<[f64; 1]> = lag_window.take(16).collect();
        println!("window autocorr: {:?}", &data);
        let window: Window<[f64; 1], Hanning> = Window::new(16);
        let mut manual: Vec<f64> = {
            let mut d: Vec<[f64; 1]> = window.take(16).collect();
            d.to_sample_slice().autocorrelate(16)
        };
        manual.normalize();
        println!("manual autocorr: {:?}", &manual);
        for i in 0..16 {
            let diff = (manual[i] - data.to_sample_slice()[i]).abs();
            assert!(diff < 1e-1);
        }
    }

    #[test]
    fn test_rms() {
        let sine = sine(64);
        let rms = sine.rms();
        println!("rms is {:?}", rms);
        assert!((rms - 0.707).abs() < 0.001);
    }
}