kord 0.8.1

A tool to easily explore music theory principles.
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

//! Generic data structures and functions for training or inference.

use burn::tensor::{backend::Backend, Tensor, TensorData};

use super::{helpers::u128_to_binary, KordItem, INPUT_SPACE_SIZE, TARGET_SPACE_SIZE};

#[cfg(feature = "ml_target_folded_bass")]
use super::PITCH_CLASS_COUNT;

#[cfg(any(feature = "ml_target_folded", feature = "ml_target_folded_bass"))]
use super::helpers::fold_binary;

#[cfg(feature = "ml_loader_note_binned_convolution")]
use super::helpers::note_binned_convolution;

#[cfg(feature = "ml_loader_mel")]
use super::helpers::mel_filter_banks_from;

#[cfg(feature = "ml_loader_frequency_pooled")]
use super::helpers::average_pool_frequency_space;

#[cfg(feature = "ml_loader_include_deterministic_guess")]
use super::{helpers::get_deterministic_guess, DETERMINISTIC_GUESS_SIZE};

/// Takes a loaded kord item and converts it to a sample tensor that is ready for classification.
pub fn kord_item_to_sample_tensor<B: Backend>(device: &B::Device, item: &KordItem) -> Tensor<B, 2> {
    #[cfg(feature = "ml_loader_note_binned_convolution")]
    {
        sample_from_note_binned_convolution(device, item)
    }

    #[cfg(feature = "ml_loader_mel")]
    {
        sample_from_mel(device, item)
    }

    #[cfg(feature = "ml_loader_frequency")]
    {
        sample_from_frequency(device, item)
    }

    #[cfg(feature = "ml_loader_frequency_pooled")]
    {
        sample_from_frequency_pooled(device, item)
    }
}

#[cfg(feature = "ml_loader_note_binned_convolution")]
fn sample_from_note_binned_convolution<B: Backend>(device: &B::Device, item: &KordItem) -> Tensor<B, 2> {
    let mut convolution = note_binned_convolution(&item.frequency_space);
    normalize(&mut convolution);
    sample_tensor_from_parts(device, item, convolution.to_vec())
}

#[cfg(feature = "ml_loader_mel")]
fn sample_from_mel<B: Backend>(device: &B::Device, item: &KordItem) -> Tensor<B, 2> {
    let mut mel_space = mel_filter_banks_from(&item.frequency_space);
    normalize(&mut mel_space);
    sample_tensor_from_parts(device, item, mel_space.to_vec())
}

#[cfg(feature = "ml_loader_frequency")]
fn sample_from_frequency<B: Backend>(device: &B::Device, item: &KordItem) -> Tensor<B, 2> {
    let mut frequency_space = item.frequency_space;
    normalize(&mut frequency_space);
    sample_tensor_from_parts(device, item, frequency_space.to_vec())
}

#[cfg(feature = "ml_loader_frequency_pooled")]
fn sample_from_frequency_pooled<B: Backend>(device: &B::Device, item: &KordItem) -> Tensor<B, 2> {
    let mut frequency_space = average_pool_frequency_space(&item.frequency_space);
    normalize(&mut frequency_space);
    sample_tensor_from_parts(device, item, frequency_space.to_vec())
}

#[allow(unused_variables)]
fn sample_tensor_from_parts<B: Backend>(device: &B::Device, item: &KordItem, mut features: Vec<f32>) -> Tensor<B, 2> {
    #[cfg(feature = "ml_loader_include_deterministic_guess")]
    {
        let mut combined = Vec::with_capacity(DETERMINISTIC_GUESS_SIZE + features.len());
        combined.extend_from_slice(&deterministic_guess_array(item));
        combined.extend_from_slice(&features);
        features = combined;
    }

    to_zero_mean_unit_variance(features.as_mut_slice());

    tensor_from_vec_with_expected_size(device, features, INPUT_SPACE_SIZE)
}

/// Takes a loaded kord item and converts it to a target tensor that is ready for classification.
#[cfg(feature = "ml_target_full")]
pub fn kord_item_to_target_tensor<B: Backend>(device: &B::Device, item: &KordItem) -> Tensor<B, 2> {
    let binary_full = u128_to_binary(item.label);
    tensor_from_vec_with_expected_size(device, binary_full.to_vec(), TARGET_SPACE_SIZE)
}

/// Takes a loaded kord item and converts it to a folded target tensor that is ready for classification.
#[cfg(feature = "ml_target_folded")]
pub fn kord_item_to_target_tensor<B: Backend>(device: &B::Device, item: &KordItem) -> Tensor<B, 2> {
    let binary_full = u128_to_binary(item.label);
    let folded = fold_binary(&binary_full);
    tensor_from_vec_with_expected_size(device, folded.to_vec(), TARGET_SPACE_SIZE)
}

/// Takes a loaded kord item and converts it to a folded+bass target tensor that is ready for classification.
#[cfg(feature = "ml_target_folded_bass")]
pub fn kord_item_to_target_tensor<B: Backend>(device: &B::Device, item: &KordItem) -> Tensor<B, 2> {
    let binary_full = u128_to_binary(item.label);
    let folded = fold_binary(&binary_full);
    let bass = lowest_pitch_class_mask(item.label);

    let mut components = Vec::with_capacity(TARGET_SPACE_SIZE);
    components.extend_from_slice(&bass);
    components.extend_from_slice(&folded);

    tensor_from_vec_with_expected_size(device, components, TARGET_SPACE_SIZE)
}

/// Modifies a slice in place to convert values to zero mean and unit variance.
pub fn to_zero_mean_unit_variance(slice: &mut [f32]) {
    let mean = slice.iter().sum::<f32>() / slice.len() as f32;
    let variance = slice.iter().map(|x| (x - mean).powf(2.0)).sum::<f32>() / slice.len() as f32;
    let std = variance.sqrt();

    if std == 0.0 {
        slice.fill(0.0);
    } else {
        slice.iter_mut().for_each(|x| *x = (*x - mean) / std);
    }
}

/// Normalizes a slice in place.
pub fn normalize(slice: &mut [f32]) {
    let max = slice.iter().fold(0f32, |acc, &x| acc.max(x));

    if max == 0.0 {
        return;
    }

    slice.iter_mut().for_each(|x| *x /= max);
}

fn tensor_from_vec_with_expected_size<B: Backend>(device: &B::Device, data: Vec<f32>, expected: usize) -> Tensor<B, 2> {
    debug_assert_eq!(data.len(), expected, "Tensor length mismatch: expected {expected}, received {}", data.len());

    let len = data.len();
    let tensor_data = TensorData::from(data.as_slice()).convert::<B::FloatElem>();
    let tensor = Tensor::<B, 1>::from_data(tensor_data, device);

    tensor.reshape([1, len])
}

#[cfg(feature = "ml_loader_include_deterministic_guess")]
fn deterministic_guess_array(item: &KordItem) -> [f32; DETERMINISTIC_GUESS_SIZE] {
    let guess_binary = get_deterministic_guess(item);
    u128_to_binary(guess_binary)
}

#[cfg(feature = "ml_target_folded_bass")]
fn lowest_pitch_class_mask(label: u128) -> [f32; PITCH_CLASS_COUNT] {
    let mut mask = [0.0; PITCH_CLASS_COUNT];

    if label != 0 {
        let lowest = label.trailing_zeros() as usize;
        let pitch_class = lowest % PITCH_CLASS_COUNT;
        mask[pitch_class] = 1.0;
    }

    mask
}

#[cfg(all(test, feature = "ml_target_folded_bass"))]
mod tests {
    use super::*;

    #[test]
    fn lowest_pitch_class_mask_marks_expected_bin() {
        let label = 1u128 << 17; // arbitrary note index 17 -> pitch class 5
        let mask = lowest_pitch_class_mask(label);

        assert_eq!(mask.iter().sum::<f32>(), 1.0);
        assert_eq!(mask[5], 1.0);
    }

    #[test]
    fn lowest_pitch_class_mask_handles_empty_label() {
        let mask = lowest_pitch_class_mask(0);
        assert!(mask.iter().all(|v| *v == 0.0));
    }
}