2 Copyright (C) 2003-2009 Paul Brossier <piem@aubio.org>
4 This file is part of aubio.
6 aubio is free software: you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation, either version 3 of the License, or
9 (at your option) any later version.
11 aubio is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with aubio. If not, see <http://www.gnu.org/licenses/>.
23 Spectral description functions
25 All of the following spectral description functions take as arguments the FFT
26 of a windowed signal (as created with aubio_pvoc). They output one smpl_t per
27 buffer (stored in a vector of size [1]).
29 A list of the spectral description methods currently available follows.
31 \section onsetdesc Onset detection functions
33 These functions are designed to raise at notes attacks in music signals.
35 \b \p energy : Energy based onset detection function
37 This function calculates the local energy of the input spectral frame.
39 \b \p hfc : High Frequency Content onset detection function
41 This method computes the High Frequency Content (HFC) of the input spectral
42 frame. The resulting function is efficient at detecting percussive onsets.
44 Paul Masri. Computer modeling of Sound for Transformation and Synthesis of
45 Musical Signal. PhD dissertation, University of Bristol, UK, 1996.
47 \b \p complex : Complex Domain Method onset detection function
49 Christopher Duxbury, Mike E. Davies, and Mark B. Sandler. Complex domain
50 onset detection for musical signals. In Proceedings of the Digital Audio
51 Effects Conference, DAFx-03, pages 90-93, London, UK, 2003.
53 \b \p phase : Phase Based Method onset detection function
55 Juan-Pablo Bello, Mike P. Davies, and Mark B. Sandler. Phase-based note onset
56 detection for music signals. In Proceedings of the IEEE International
57 Conference on Acoustics Speech and Signal Processing, pages 441444,
60 \b \p specdiff : Spectral difference method onset detection function
62 Jonhatan Foote and Shingo Uchihashi. The beat spectrum: a new approach to
63 rhythm analysis. In IEEE International Conference on Multimedia and Expo
64 (ICME 2001), pages 881884, Tokyo, Japan, August 2001.
66 \b \p kl : Kullback-Liebler onset detection function
68 Stephen Hainsworth and Malcom Macleod. Onset detection in music audio
69 signals. In Proceedings of the International Computer Music Conference
70 (ICMC), Singapore, 2003.
72 \b \p mkl : Modified Kullback-Liebler onset detection function
74 Paul Brossier, ``Automatic annotation of musical audio for interactive
75 systems'', Chapter 2, Temporal segmentation, PhD thesis, Centre for Digital
76 music, Queen Mary University of London, London, UK, 2006.
78 \b \p specflux : Spectral Flux
80 Simon Dixon, Onset Detection Revisited, in ``Proceedings of the 9th
81 International Conference on Digital Audio Effects'' (DAFx-06), Montreal,
84 \section shapedesc Spectral shape descriptors
86 The following descriptors are described in:
88 Geoffroy Peeters, <i>A large set of audio features for sound description
89 (similarity and classification) in the CUIDADO project</i>, CUIDADO I.S.T.
90 Project Report 2004 (<a
91 href="http://www.ircam.fr/anasyn/peeters/ARTICLES/Peeters_2003_cuidadoaudiofeatures.pdf">pdf</a>)
93 \b \p centroid : Spectral centroid
95 The spectral centroid represents the barycenter of the spectrum.
97 \e Note: This function returns the result in bin. To get the spectral
98 centroid in Hz, aubio_bintofreq() should be used.
100 \b \p spread : Spectral spread
102 The spectral spread is the variance of the spectral distribution around its
105 See also <a href="http://en.wikipedia.org/wiki/Standard_deviation">Standard
106 deviation</a> on Wikipedia.
108 \b \p skewness : Spectral skewness
110 Similarly, the skewness is computed from the third order moment of the
111 spectrum. A negative skewness indicates more energy on the lower part of the
112 spectrum. A positive skewness indicates more energy on the high frequency of
115 See also <a href="http://en.wikipedia.org/wiki/Skewness">Skewness</a> on
118 \b \p kurtosis : Spectral kurtosis
120 The kurtosis is a measure of the flatness of the spectrum, computed from the
123 See also <a href="http://en.wikipedia.org/wiki/Kurtosis">Kurtosis</a> on
126 \b \p slope : Spectral slope
128 The spectral slope represents decreasing rate of the spectral amplitude,
129 computed using a linear regression.
131 \b \p decrease : Spectral decrease
133 The spectral decrease is another representation of the decreasing rate,
134 based on perceptual criteria.
136 \b \p rolloff : Spectral roll-off
138 This function returns the bin number below which 95% of the spectrum energy
144 #ifndef ONSETDETECTION_H
145 #define ONSETDETECTION_H
151 /** spectral description structure */
152 typedef struct _aubio_specdesc_t aubio_specdesc_t;
154 /** execute spectral description function on a spectral frame
156 Generic function to compute spectral detescription.
158 \param o spectral description object as returned by new_aubio_specdesc()
159 \param fftgrain input signal spectrum as computed by aubio_pvoc_do
160 \param desc output vector (one sample long, to send to the peak picking)
163 void aubio_specdesc_do (aubio_specdesc_t * o, cvec_t * fftgrain,
166 /** creation of a spectral description object
168 \param method spectral description method
169 \param buf_size length of the input spectrum frame
172 aubio_specdesc_t *new_aubio_specdesc (char_t * method, uint_t buf_size);
174 /** deletion of a spectral descriptor
176 \param o spectral descriptor object as returned by new_aubio_specdesc()
179 void del_aubio_specdesc (aubio_specdesc_t * o);
185 #endif /* ONSETDETECTION_H */