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
/* MIT License Copyright (c) 2021 Philipp Schuster Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ //! A simple and fast `no_std` library to get the frequency spectrum of a digital signal //! (e.g. audio) using FFT. It follows the KISS principle and consists of simple building //! blocks/optional features. #![no_std] // use alloc crate, because this is no_std // #[macro_use] extern crate alloc; // use std in tests #[cfg(test)] #[macro_use] extern crate std; use alloc::vec::Vec; use rustfft::algorithm::Radix4; use rustfft::num_complex::Complex32; use rustfft::{Fft, FftDirection}; pub use crate::frequency::{Frequency, FrequencyValue}; pub use crate::limit::FrequencyLimit; pub use crate::spectrum::{FrequencySpectrum, SpectrumTotalScaleFunctionFactory}; use core::convert::identity; mod frequency; mod limit; mod spectrum; #[cfg(test)] mod tests; pub mod windows; /// Takes an array of samples (length must be a power of 2), /// e.g. 2048, applies an FFT (using library `rustfft`) on it /// and returns all frequencies with their volume/magnitude. /// /// * `samples` raw audio, e.g. 16bit audio data but as f32. /// You should apply an window function (like Hann) on the data first. /// The final frequency resolution is `sample_rate / (N / 2)` /// e.g. `44100/(16384/2) == 5.383Hz`, i.e. more samples => better accuracy /// * `sampling_rate` sampling_rate, e.g. `44100 [Hz]` /// * `frequency_limit` Frequency limit. See [`FrequencyLimit´] /// * `per_element_scaling_fn` Optional per element scaling function, e.g. `20 * log(x)`. /// To see where this equation comes from, check out /// this paper: /// https://www.sjsu.edu/people/burford.furman/docs/me120/FFT_tutorial_NI.pdf /// * `total_scaling_fn` See [`crate::spectrum::SpectrumTotalScaleFunctionFactory`]. /// /// ## Returns value /// New object of type [`FrequencySpectrum`]. pub fn samples_fft_to_spectrum( samples: &[f32], sampling_rate: u32, frequency_limit: FrequencyLimit, per_element_scaling_fn: Option<&dyn Fn(f32) -> f32>, total_scaling_fn: Option<SpectrumTotalScaleFunctionFactory>, ) -> FrequencySpectrum { // With FFT we transform an array of time-domain waveform samples // into an array of frequency-domain spectrum samples // https://www.youtube.com/watch?v=z7X6jgFnB6Y // FFT result has same length as input // convert to Complex for FFT let mut buffer = samples_to_complex(samples); // a power of 2, like 1024 or 2048 let fft_len = samples.len(); // apply the fft let fft = Radix4::new(fft_len, FftDirection::Forward); fft.process(&mut buffer); // we only need the first half of the results with FFT // because of Nyquist theorem. 44100hz sampling frequency // => 22050hz maximum detectable frequency // This function: // 1) calculates the corresponding frequency of each index in the FFT result // 2) filters out unwanted frequencies // 3) calculates the magnitude (absolute value) at each frequency index for each complex value // 4) optionally scales the magnitudes // 5) collects everything into the struct "FrequencySpectrum" fft_result_to_frequency_to_magnitude_map( buffer, sampling_rate, frequency_limit, per_element_scaling_fn, total_scaling_fn, ) } /// Converts all samples to a complex number (imaginary part is set to two) /// as preparation for the FFT. /// /// ## Parameters /// `samples` Input samples. /// /// ## Return value /// New vector of samples but as Complex data type. #[inline(always)] fn samples_to_complex(samples: &[f32]) -> Vec<Complex32> { samples .iter() .map(|x| Complex32::new(x.clone(), 0.0)) .collect::<Vec<Complex32>>() } /// Transforms the complex numbers of the first half of the FFT results (only the first /// half is relevant, Nyquist theorem) to their magnitudes. /// /// ## Parameters /// * `fft_result` Result buffer from FFT. Has the same length as the samples array. /// * `sampling_rate` sampling_rate, e.g. `44100 [Hz]` /// * `frequency_limit` Frequency limit. See [`FrequencyLimit´] /// * `per_element_scaling_fn` Optional per element scaling function, e.g. `20 * log(x)`. /// To see where this equation comes from, check out /// this paper: /// https://www.sjsu.edu/people/burford.furman/docs/me120/FFT_tutorial_NI.pdf /// * `total_scaling_fn` See [`crate::spectrum::SpectrumTotalScaleFunctionFactory`]. /// /// ## Return value /// New object of type [`FrequencySpectrum`]. #[inline(always)] fn fft_result_to_frequency_to_magnitude_map( fft_result: Vec<Complex32>, sampling_rate: u32, frequency_limit: FrequencyLimit, per_element_scaling_fn: Option<&dyn Fn(f32) -> f32>, total_scaling_fn: Option<SpectrumTotalScaleFunctionFactory>, ) -> FrequencySpectrum { let maybe_min = frequency_limit.maybe_min(); let maybe_max = frequency_limit.maybe_max(); let samples_len = fft_result.len(); // collect frequency => frequency value in Vector of Pairs/Tuples let frequency_vec = fft_result .into_iter() // take first half; half of input length .take(samples_len / 2) // get (index, complex)-pairs .enumerate() // calc index => corresponding frequency .map(|(fft_index, complex)| { ( fft_index_to_corresponding_frequency(fft_index, samples_len as u32, sampling_rate), complex, ) }) // ####################### // ### BEGIN filtering: results in lower calculation and memory overhead! // check lower bound frequency (inclusive) .filter(|(fr, _complex)| { if let Some(min_fr) = maybe_min { // inclusive! *fr >= min_fr } else { true } }) // check upper bound frequency (inclusive) .filter(|(fr, _complex)| { if let Some(max_fr) = maybe_max { // inclusive! *fr <= max_fr } else { true } }) // ### END filtering // ####################### // calc magnitude: sqrt(re*re + im*im) (re: real part, im: imaginary part) .map(|(fr, complex)| (fr, complex.norm())) // apply optionally scale function .map(|(fr, val)| (fr, per_element_scaling_fn.unwrap_or(&identity)(val))) // transform to my thin convenient orderable f32 wrappers .map(|(fr, val)| (Frequency::from(fr), FrequencyValue::from(val))) .collect::<Vec<(Frequency, FrequencyValue)>>(); // create spectrum object let fs = FrequencySpectrum::new(frequency_vec); // optionally scale if let Some(total_scaling_fn) = total_scaling_fn { fs.apply_total_scaling_fn(total_scaling_fn) } fs } /// Calculate what index in the FFT result corresponds to what frequency. /// /// ## Parameters /// * `fft_index` Index in FFT result buffer. If `samples.len() == 2048` then this is in `{0, 1, ..., 1023}` /// * `samples_len` Number of samples put into the FFT /// * `sampling_rate` sampling_rate, e.g. `44100 [Hz]` /// /// ## Return value #[inline(always)] fn fft_index_to_corresponding_frequency( fft_index: usize, samples_len: u32, sampling_rate: u32, ) -> f32 { // Explanation for the algorithm: // https://stackoverflow.com/questions/4364823/ // samples : [0], [1], [2], [3], ... , ..., [2047] => 2048 samples for example // FFT Result : [0], [1], [2], [3], ... , ..., [2047] // Relevant part of FFT Result: [0], [1], [2], [3], ... , [1023] // ^ ^ // Frequency : 0Hz, .................... Sampling Rate/2 // 0Hz is also called (e.g. 22050Hz @ 44100H sampling rate) // "DC Component" // frequency step/resolution is for example: 1/1024 * 44100 // 1024: relevant FFT result, 2048 samples, 44100 sample rate fft_index as f32 / samples_len as f32 * sampling_rate as f32 }