1AUBIOPITCH(1) aubio User's manual AUBIOPITCH(1)
2
6 aubiopitch - a command line tool to extract musical pitch
7
9 aubiopitch source
10 aubiopitch [[-i] source] [-o sink]
11 [-r rate] [-B win] [-H hop]
12 [-p method] [-u unit] [-l thres]
13 [-T time-format]
14 [-s sil] [-f]
15 [-v] [-h] [-j]
16
19 aubiopitch attempts to detect the pitch, the perceived height of a mu‐
20 sical note.
21 When started with an input source (-i/--input), the detected pitch are
23 printed on the console, prefixed by a timestamp in seconds. If no pitch
24 candidate is found, the output is 0.
25 When started without an input source, or with the jack option
27 (-j/--jack), aubiopitch starts in jack mode.
28
30 This program follows the usual GNU command line syntax, with long op‐
31 tions starting with two dashes (--). A summary of options is included
32 below.
33 -i, --input source
35 Run analysis on this audio file. Most uncompressed and com‐
36 pressed are supported, depending on how aubio was built.
37 -o, --output sink
39 Save results in this file. The file will be created on the model
40 of the input file. The detected frequency is played at the de‐
41 tected loudness.
42 -r, --samplerate rate
44 Fetch the input source, resampled at the given sampling rate.
45 The rate should be specified in Hertz as an integer. If 0, the
46 sampling rate of the original source will be used. Defaults to
47 0.
48 -B, --bufsize win
50 The size of the buffer to analyze, that is the length of the
51 window used for spectral and temporal computations. Defaults to
52 2048.
53 -H, --hopsize hop
55 The number of samples between two consecutive analysis. De‐
56 faults to 256.
57 -p, --pitch method
59 The pitch detection method to use. See PITCH METHODS below. De‐
60 faults to 'default'.
61 -u, --pitch-unit unit
63 The unit to be used to print frequencies. Possible values in‐
64 clude midi, bin, cent, and Hz. Defaults to 'Hz'.
65 -l, --pitch-tolerance thres
67 Set the tolerance for the pitch detection algorithm. Typical
68 values range between 0.2 and 0.9. Pitch candidates found with a
69 confidence less than this threshold will not be selected. The
70 higher the threshold, the more confidence in the candidates. De‐
71 faults to unset.
72 -s, --silence sil
74 Set the silence threshold, in dB, under which the onset will not
75 be detected. A value of -20.0 would eliminate most onsets but
76 the loudest ones. A value of -90.0 would select all onsets. De‐
77 faults to -90.0.
78 -T, --timeformat format
80 Set time format (samples, ms, seconds). Defaults to seconds.
81 -m, --mix-input
83 Mix source signal to the output signal before writing to sink.
84 -f, --force-overwrite
86 Overwrite output file if it already exists.
87 -j, --jack
89 Use Jack input/output. You will need a Jack connection con‐
90 troller to feed aubio some signal and listen to its output.
91 -h, --help
93 Print a short help message and exit.
94 -v, --verbose
96 Be verbose.
97
99 Available methods are:
100 default
102 use the default method
103 Currently, the default method is set to yinfft.
105 schmitt
107 Schmitt trigger
108 This pitch extraction method implements a Schmitt trigger to estimate
110 the period of a signal. It is computationally very inexpensive, but
111 also very sensitive to noise.
112 fcomb a fast harmonic comb filter
114 This pitch extraction method implements a fast harmonic comb filter to
116 determine the fundamental frequency of a harmonic sound.
117 mcomb multiple-comb filter
119 This fundamental frequency estimation algorithm implements spectral
121 flattening, multi-comb filtering and peak histogramming.
122 specacf
124 Spectral auto-correlation function
125 yin YIN algorithm
127 This algorithm was developed by A. de Cheveigne and H. Kawahara and was
129 first published in:
130 De Cheveigné, A., Kawahara, H. (2002) "YIN, a fundamental frequency es‐
132 timator for speech and music", J. Acoust. Soc. Am. 111, 1917-1930.
133 yinfft Yinfft algorithm
135 This algorithm was derived from the YIN algorithm. In this implementa‐
137 tion, a Fourier transform is used to compute a tapered square differ‐
138 ence function, which allows spectral weighting. Because the difference
139 function is tapered, the selection of the period is simplified.
140 Paul Brossier, Automatic annotation of musical audio for interactive
142 systems, Chapter 3, Pitch Analysis, PhD thesis, Centre for Digital mu‐
143 sic, Queen Mary University of London, London, UK, 2006.
144 yinfast
146 YIN algorithm (accelerated)
147 An optimised implementation of the YIN algorithm, yielding results
149 identical to the original YIN algorithm, while reducing its computa‐
150 tional cost from O(n^2) to O(n log(n)).
151
153 aubioonset(1), aubiotrack(1), aubionotes(1), aubioquiet(1), aubiom‐
154 fcc(1), and aubiocut(1).
155
157 This manual page was written by Paul Brossier <piem@aubio.org>. Permis‐
158 sion is granted to copy, distribute and/or modify this document under
159 the terms of the GNU General Public License as published by the Free
160 Software Foundation, either version 3 of the License, or (at your op‐
161 tion) any later version.
162aubio 0.4.9 21 July 2021 AUBIOPITCH(1)