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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282
/*
* Copyright (C) Simon Werner, 2022.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, see <http://www.gnu.org/licenses/>.
*/
use std::f32;
#[cfg(feature = "png")]
use std::path::Path;
use crate::errors::SonogramError;
use crate::window_fn;
use crate::SpecCompute;
type WindowFn = fn(usize, usize) -> f32;
///
/// A builder struct that will output a spectrogram creator when complete.
/// This builder will require the height and width of the final spectrogram,
/// at a minimum. However you can load data from a .wav file, or directly
/// from a Vec<i16> memory object.
///
/// # Example
///
/// ```Rust
/// let mut spectrograph = SpecOptionsBuilder::new(512, 128)
/// .set_window_fn(utility::blackman_harris)
/// .load_data_from_file(&std::path::Path::new("test.wav"))?
/// .build();
/// ```
///
pub struct SpecOptionsBuilder {
// Inputs
data: Vec<f32>, // Our time-domain data (audio samples)
sample_rate: u32, // The sample rate of the wav data
channel: u16, // The audio channel
scale_factor: Option<f32>, // How much to scale the sample amplitude by
do_normalise: bool, // Normalise the samples to between -1.0...1.0
downsample_divisor: Option<usize>, // Downsample the samples by a given amount
// FFT info
num_bins: usize, // The number of FFT bins
step_size: usize, // How far to step between each window function
window_fn: WindowFn, // The windowing function to use.
}
impl SpecOptionsBuilder {
/// Create a new SpecOptionsBuilder. The final height and width of
/// the spectrogram must be supplied. Before the `build` function
/// can be called a `load_data_from_*` function needs to be called.
///
/// # Arguments
///
/// * `num_bins` - Number of bins in the discrete fourier transform (FFT)
///
pub fn new(num_bins: usize) -> Self {
SpecOptionsBuilder {
data: vec![],
sample_rate: 11025,
channel: 1,
scale_factor: None,
do_normalise: false,
downsample_divisor: None,
num_bins,
window_fn: window_fn::rectangular,
step_size: num_bins,
}
}
/// Load a .wav file to memory and use that file as the input.
///
/// # Arguments
///
/// * `fname` - The path to the file.
///
#[cfg(feature = "hound")]
pub fn load_data_from_file(self, fname: &Path) -> Result<Self, SonogramError> {
let mut reader = hound::WavReader::open(fname)?;
// Can only handle 16 bit data
// TODO: Add more data here
if 16 != reader.spec().bits_per_sample {
return Err(SonogramError::InvalidCodec);
}
if self.channel > reader.spec().channels {
return Err(SonogramError::InvalidChannel);
}
let data: Vec<i16> = {
let first_sample = self.channel as usize - 1;
let step_size = reader.spec().channels as usize;
let mut s = reader.samples();
// TODO: replace this with .advanced_by in the future
for _ in 0..first_sample {
s.next();
}
s.step_by(step_size).map(|x| x.unwrap()).collect()
};
let sample_rate = reader.spec().sample_rate;
Ok(self.load_data_from_memory(data, sample_rate))
}
/// Load data directly from memory - i16 version.
///
/// # Arguments
///
/// * `data` - The raw wavform data that will be converted to a spectrogram.
/// * `sample_rate` - The sample rate, in Hz, of the data.
///
pub fn load_data_from_memory(mut self, data: Vec<i16>, sample_rate: u32) -> Self {
self.data = data.iter().map(|&x| x as f32 / (i16::MAX as f32)).collect();
self.sample_rate = sample_rate;
self
}
/// Load data directly from memory - f32 version.
///
/// # Arguments
///
/// * `data` - The raw wavform data that will be converted to a spectrogram.
/// Samples must be in the range -1.0 to 1.0.
/// * `sample_rate` - The sample rate, in Hz, of the data.
///
pub fn load_data_from_memory_f32(mut self, data: Vec<f32>, sample_rate: u32) -> Self {
self.data = data;
self.sample_rate = sample_rate;
self
}
///
/// Down sample the data by the given divisor. This is a cheap way of
/// improving the performance of the FFT.
///
/// # Arguments
///
/// * `divisor` - How much to reduce the data by.
///
pub fn downsample(mut self, divisor: usize) -> Self {
self.downsample_divisor = Some(divisor);
self
}
///
/// Set the audio channel to use when importing a WAV file.
/// By default this is 1.
///
pub fn channel(mut self, channel: u16) -> Self {
self.channel = channel;
self
}
///
/// Normalise all the sample values to range from -1.0 to 1.0.
///
pub fn normalise(mut self) -> Self {
self.do_normalise = true;
self
}
///
/// Scale the sample data by the given amount.
///
pub fn scale(mut self, scale_factor: f32) -> Self {
self.scale_factor = Some(scale_factor);
self
}
/// A window function describes the type of window to use during the
/// DFT (discrete fourier transform). See
/// (here)[https://en.wikipedia.org/wiki/Window_function] for more details.
///
/// # Arguments
///
/// * `window` - The window function to be used.
///
pub fn set_window_fn(mut self, window_fn: WindowFn) -> Self {
self.window_fn = window_fn;
self
}
///
/// This is the step size (as the number of samples) between each
/// application of the window function. A smaller step size may
/// increase the smoothness of the sample, but take more time. The default
/// step size, if not set, is the same as the number of FFT bins. This
/// there is no overlap between windows and it most cases will suit your
/// needs.
///
pub fn set_step_size(mut self, step_size: usize) -> Self {
self.step_size = step_size;
self
}
///
/// The final method to be called. This will create an instance of
/// [Spectrograph].
///
pub fn build(mut self) -> Result<SpecCompute, SonogramError> {
if self.data.is_empty() {
// SpecOptionsBuilder requires data to be loaded
return Err(SonogramError::IncompleteData);
}
if self.channel == 0 {
// The channel must be an integer 1 or greater
return Err(SonogramError::InvalidChannel);
}
//
// Do downsample
//
if let Some(divisor) = self.downsample_divisor {
if divisor == 0 {
return Err(SonogramError::InvalidDivisor);
}
if divisor > 1 {
for (j, i) in (0..self.data.len() - divisor).step_by(divisor).enumerate() {
let sum: f32 = self.data[i..i + divisor].iter().fold(0.0, |mut sum, &val| {
sum += val;
sum
});
let avg = sum / (divisor as f32);
self.data[j] = avg;
}
self.data.resize(self.data.len() / divisor, 0.0);
self.sample_rate /= divisor as u32;
}
}
//
// Normalise
//
if self.do_normalise {
let max = self
.data
.iter()
.reduce(|max, x| if x > max { x } else { max })
.unwrap();
let norm = 1.0 / max;
for x in self.data.iter_mut() {
*x *= norm;
}
}
//
// Apply the scale factor
//
if let Some(scale_factor) = self.scale_factor {
for x in self.data.iter_mut() {
*x *= scale_factor;
}
}
Ok(SpecCompute::new(
self.num_bins,
self.step_size,
self.data,
self.window_fn,
))
}
}