Command line tools¶
A few simple command line tools are included along with the library.
aubioonsetoutputs the time stamp of detected note onsetsaubiopitchattempts to identify a fundamental frequency, or pitch, for each frame of the input soundaubiomfcccomputes Mel-frequency Cepstrum Coefficientsaubiotrackoutputs the time stamp of detected beatsaubionotesemits midi-like notes, with an onset, a pitch, and a durationaubioquietextracts quiet and loud regions
Additionally, the python module comes with the following script:
aubiocutslices sound files at onset or beat timestamps
aubioonset¶
NAME
aubioonset - a command line tool to extract musical onset times
SYNOPSIS
aubioonset source
aubioonset [[-i] source] [-o sink]
[-r rate] [-B win] [-H hop]
[-O method] [-t thres]
[-T time-format]
[-s sil] [-m] [-f]
[-j] [-v] [-h]
DESCRIPTION
aubioonset attempts to detect onset times, the beginning of discrete sound
events, in audio signals.
When started with an input source (-i/--input), the detected onset times are
given on the console, in seconds.
When started without an input source, or with the jack option (-j/--jack),
aubioonset starts in jack mode.
OPTIONS
This program follows the usual GNU command line syntax, with long options
starting with two dashes (--). A summary of options is included below.
-i, --input source Run analysis on this audio file. Most uncompressed and
compressed are supported, depending on how aubio was built.
-o, --output sink Save results in this file. The file will be created on
the model of the input file. Onset times are marked by a short wood-block
like sound.
-r, --samplerate rate Fetch the input source, resampled at the given
sampling rate. The rate should be specified in Hertz as an integer. If 0,
the sampling rate of the original source will be used. Defaults to 0.
-B, --bufsize win The size of the buffer to analyze, that is the length
of the window used for spectral and temporal computations. Defaults to 512.
-H, --hopsize hop The number of samples between two consecutive analysis.
Defaults to 256.
-O, --onset method The onset detection method to use. See ONSET METHODS
below. Defaults to 'default'.
-t, --onset-threshold thres Set the threshold value for the onset peak
picking. Values are typically in the range [0.001, 0.900]. Lower threshold
values imply more onsets detected. Increasing this threshold should reduce
the number of incorrect detections. Defaults to 0.3.
-M, --minioi value Set the minimum inter-onset interval, in seconds, the
shortest interval between two consecutive onsets. Defaults to 0.020
-s, --silence sil Set the silence threshold, in dB, under which the onset
will not be detected. A value of -20.0 would eliminate most onsets but the
loudest ones. A value of -90.0 would select all onsets. Defaults to -90.0.
-T, --timeformat format Set time format (samples, ms, seconds). Defaults to
seconds.
-m, --mix-input Mix source signal to the output signal before writing to
sink.
-f, --force-overwrite Overwrite output file if it already exists.
-j, --jack Use Jack input/output. You will need a Jack connection
controller to feed aubio some signal and listen to its output.
-h, --help Print a short help message and exit.
-v, --verbose Be verbose.
ONSET METHODS
Available methods are:
default Default distance, currently hfc
Default: 'default' (currently set to hfc)
energy Energy based distance
This function calculates the local energy of the input spectral frame.
hfc High-Frequency content
This method computes the High Frequency Content (HFC) of the input
spectral frame. The resulting function is efficient at detecting
percussive onsets.
Paul Masri. Computer modeling of Sound for Transformation and Synthesis of
Musical Signal. PhD dissertation, University of Bristol, UK, 1996.
complex Complex domain onset detection function
This function uses information both in frequency and in phase to determine
changes in the spectral content that might correspond to musical onsets.
It is best suited for complex signals such as polyphonic recordings.
Christopher Duxbury, Mike E. Davies, and Mark B. Sandler. Complex domain
onset detection for musical signals. In Proceedings of the Digital Audio
Effects Conference, DAFx-03, pages 90-93, London, UK, 2003.
phase Phase based onset detection function
This function uses information both in frequency and in phase to determine
changes in the spectral content that might correspond to musical onsets. It
is best suited for complex signals such as polyphonic recordings.
Juan-Pablo Bello, Mike P. Davies, and Mark B. Sandler. Phase-based note
onset detection for music signals. In Proceedings of the IEEE International
Conference on Acoustics Speech and Signal Processing, pages 441444,
Hong-Kong, 2003.
specdiff Spectral difference onset detection function
Jonhatan Foote and Shingo Uchihashi. The beat spectrum: a new approach to
rhythm analysis. In IEEE International Conference on Multimedia and Expo
(ICME 2001), pages 881884, Tokyo, Japan, August 2001.
kl Kulback-Liebler onset detection function
Stephen Hainsworth and Malcom Macleod. Onset detection in music audio
signals. In Proceedings of the International Computer Music Conference
(ICMC), Singapore, 2003.
mkl Modified Kulback-Liebler onset detection function
Paul Brossier, ``Automatic annotation of musical audio for interactive
systems'', Chapter 2, Temporal segmentation, PhD thesis, Centre for
Digital music, Queen Mary University of London, London, UK, 2006.
specflux Spectral flux
Simon Dixon, Onset Detection Revisited, in ``Proceedings of the 9th
International Conference on Digital Audio Effects'' (DAFx-06), Montreal,
Canada, 2006.
SEE ALSO
aubiopitch(1),
aubiotrack(1),
aubionotes(1),
aubioquiet(1),
aubiomfcc(1),
and
aubiocut(1).
AUTHOR
This manual page was written by Paul Brossier <piem@aubio.org>. Permission is
granted to copy, distribute and/or modify this document under the terms of
the GNU General Public License as published by the Free Software Foundation,
either version 3 of the License, or (at your option) any later version.
aubiopitch¶
NAME
aubiopitch - a command line tool to extract musical pitch
SYNOPSIS
aubiopitch source
aubiopitch [[-i] source] [-o sink]
[-r rate] [-B win] [-H hop]
[-p method] [-u unit] [-l thres]
[-T time-format]
[-s sil] [-f]
[-v] [-h] [-j]
DESCRIPTION
aubiopitch attempts to detect the pitch, the perceived height of a musical
note.
When started with an input source (-i/--input), the detected pitch are
printed on the console, prefixed by a timestamp in seconds. If no pitch
candidate is found, the output is 0.
When started without an input source, or with the jack option (-j/--jack),
aubiopitch starts in jack mode.
OPTIONS
This program follows the usual GNU command line syntax, with long options
starting with two dashes (--). A summary of options is included below.
-i, --input source Run analysis on this audio file. Most uncompressed and
compressed are supported, depending on how aubio was built.
-o, --output sink Save results in this file. The file will be created on
the model of the input file. The detected frequency is played at the
detected loudness.
-r, --samplerate rate Fetch the input source, resampled at the given
sampling rate. The rate should be specified in Hertz as an integer. If 0,
the sampling rate of the original source will be used. Defaults to 0.
-B, --bufsize win The size of the buffer to analyze, that is the length
of the window used for spectral and temporal computations. Defaults to 2048.
-H, --hopsize hop The number of samples between two consecutive analysis.
Defaults to 256.
-p, --pitch method The pitch detection method to use. See PITCH METHODS
below. Defaults to 'default'.
-u, --pitch-unit unit The unit to be used to print frequencies. Possible
values include midi, bin, cent, and Hz. Defaults to 'Hz'.
-l, --pitch-tolerance thres Set the tolerance for the pitch detection
algorithm. Typical values range between 0.2 and 0.9. Pitch candidates found
with a confidence less than this threshold will not be selected. The higher
the threshold, the more confidence in the candidates. Defaults to unset.
-s, --silence sil Set the silence threshold, in dB, under which the onset
will not be detected. A value of -20.0 would eliminate most onsets but the
loudest ones. A value of -90.0 would select all onsets. Defaults to -90.0.
-T, --timeformat format Set time format (samples, ms, seconds). Defaults to
seconds.
-m, --mix-input Mix source signal to the output signal before writing to
sink.
-f, --force-overwrite Overwrite output file if it already exists.
-j, --jack Use Jack input/output. You will need a Jack connection
controller to feed aubio some signal and listen to its output.
-h, --help Print a short help message and exit.
-v, --verbose Be verbose.
PITCH METHODS
Available methods are:
default use the default method
Currently, the default method is set to yinfft.
schmitt Schmitt trigger
This pitch extraction method implements a Schmitt trigger to estimate the
period of a signal. It is computationally very inexpensive, but also very
sensitive to noise.
fcomb a fast harmonic comb filter
This pitch extraction method implements a fast harmonic comb filter to
determine the fundamental frequency of a harmonic sound.
mcomb multiple-comb filter
This fundamental frequency estimation algorithm implements spectral
flattening, multi-comb filtering and peak histogramming.
specacf Spectral auto-correlation function
yin YIN algorithm
This algorithm was developed by A. de Cheveigne and H. Kawahara and
was first published in:
De Cheveigné, A., Kawahara, H. (2002) "YIN, a fundamental frequency
estimator for speech and music", J. Acoust. Soc. Am. 111, 1917-1930.
yinfft Yinfft algorithm
This algorithm was derived from the YIN algorithm. In this implementation, a
Fourier transform is used to compute a tapered square difference function,
which allows spectral weighting. Because the difference function is tapered,
the selection of the period is simplified.
Paul Brossier, Automatic annotation of musical audio for interactive systems,
Chapter 3, Pitch Analysis, PhD thesis, Centre for Digital music, Queen Mary
University of London, London, UK, 2006.
SEE ALSO
aubioonset(1),
aubiotrack(1),
aubionotes(1),
aubioquiet(1),
aubiomfcc(1),
and
aubiocut(1).
AUTHOR
This manual page was written by Paul Brossier <piem@aubio.org>. Permission is
granted to copy, distribute and/or modify this document under the terms of
the GNU General Public License as published by the Free Software Foundation,
either version 3 of the License, or (at your option) any later version.
aubiomfcc¶
NAME
aubiomfcc - a command line tool to compute Mel-Frequency Cepstrum Coefficients
SYNOPSIS
aubiomfcc source
aubiomfcc [[-i] source]
[-r rate] [-B win] [-H hop]
[-T time-format]
[-v] [-h]
DESCRIPTION
aubiomfcc compute the Mel-Frequency Cepstrum Coefficients (MFCC).
MFCCs are coefficients that make up for the mel-frequency spectrum, a
representation of the short-term power spectrum of a sound. By default, 13
coefficents are computed using 40 filters.
When started with an input source (-i/--input), the coefficients are given on
the console, prefixed by their timestamps in seconds.
OPTIONS
This program follows the usual GNU command line syntax, with long options
starting with two dashes (--). A summary of options is included below.
-i, --input source Run analysis on this audio file. Most uncompressed and
compressed are supported, depending on how aubio was built.
-r, --samplerate rate Fetch the input source, resampled at the given
sampling rate. The rate should be specified in Hertz as an integer. If 0,
the sampling rate of the original source will be used. Defaults to 0.
-B, --bufsize win The size of the buffer to analyze, that is the length
of the window used for spectral and temporal computations. Defaults to 512.
-H, --hopsize hop The number of samples between two consecutive analysis.
Defaults to 256.
-T, --timeformat format Set time format (samples, ms, seconds). Defaults to
seconds.
-h, --help Print a short help message and exit.
-v, --verbose Be verbose.
REFERENCES
Using the default parameters, the filter coefficients will be computed
according to Malcolm Slaney's Auditory Toolbox, available at the following
url:
http://cobweb.ecn.purdue.edu/~malcolm/interval/1998-010/ (see file mfcc.m)
SEE ALSO
aubioonset(1),
aubiopitch(1),
aubiotrack(1),
aubionotes(1),
aubioquiet(1),
and
aubiocut(1).
AUTHOR
This manual page was written by Paul Brossier <piem@aubio.org>. Permission is
granted to copy, distribute and/or modify this document under the terms of
the GNU General Public License as published by the Free Software Foundation,
either version 3 of the License, or (at your option) any later version.
aubiotrack¶
NAME
aubiotrack - a command line tool to extract musical beats from audio signals
SYNOPSIS
aubiotrack source
aubiotrack [[-i] source] [-o sink]
[-r rate] [-B win] [-H hop]
[-T time-format]
[-s sil] [-m]
[-j] [-v] [-h]
DESCRIPTION
aubiotrack attempts to detect beats, the time where one would intuitively be
tapping his foot.
When started with an input source (-i/--input), the detected beats are given
on the console, in seconds.
When started without an input source, or with the jack option (-j/--jack),
aubiotrack starts in jack mode.
OPTIONS
This program follows the usual GNU command line syntax, with long options
starting with two dashes (--). A summary of options is included below.
-i, --input source Run analysis on this audio file. Most uncompressed and
compressed are supported, depending on how aubio was built.
-o, --output sink Save results in this file. The file will be created on the
model of the input file. Beats are marked by a short wood-block like sound.
-r, --samplerate rate Fetch the input source, resampled at the given
sampling rate. The rate should be specified in Hertz as an integer. If 0,
the sampling rate of the original source will be used. Defaults to 0.
-B, --bufsize win The size of the buffer to analyze, that is the length
of the window used for spectral and temporal computations. Defaults to 512.
-H, --hopsize hop The number of samples between two consecutive analysis.
Defaults to 256.
-s, --silence sil Set the silence threshold, in dB, under which the pitch
will not be detected. A value of -20.0 would eliminate most onsets but the
loudest ones. A value of -90.0 would select all onsets. Defaults to -90.0.
-m, --mix-input Mix source signal to the output signal before writing to
sink.
-f, --force-overwrite Overwrite output file if it already exists.
-j, --jack Use Jack input/output. You will need a Jack connection
controller to feed aubio some signal and listen to its output.
-T, --timeformat format Set time format (samples, ms, seconds). Defaults to
seconds.
-h, --help Print a short help message and exit.
-v, --verbose Be verbose.
BEAT TRACKING METHODS
Aubio currently implements one the causal beat tracking algorithm designed by
Matthew Davies and described in the following articles:
Matthew E. P. Davies and Mark D. Plumbley. Causal tempo tracking of audio.
In Proceedings of the International Symposium on Music Information Retrieval
(ISMIR), pages 164169, Barcelona, Spain, 2004.
Matthew E. P. Davies, Paul Brossier, and Mark D. Plumbley. Beat tracking
towards automatic musical accompaniment. In Proceedings of the Audio
Engeeniring Society 118th Convention, Barcelona, Spain, May 2005.
SEE ALSO
aubioonset(1),
aubiopitch(1),
aubionotes(1),
aubioquiet(1),
aubiomfcc(1),
and
aubiocut(1).
AUTHOR
This manual page was written by Paul Brossier <piem@aubio.org>. Permission is
granted to copy, distribute and/or modify this document under the terms of
the GNU General Public License as published by the Free Software Foundation,
either version 3 of the License, or (at your option) any later version.
aubionotes¶
NAME
aubionotes - a command line tool to extract musical notes
SYNOPSIS
aubionotes source
aubionotes [[-i] source]
[-r rate] [-B win] [-H hop]
[-O method] [-t thres]
[-p method] [-u unit] [-l thres]
[-T time-format]
[-s sil]
[-j] [-v] [-h]
DESCRIPTION
aubionotes attempts to detect notes by looking for note onsets and pitches.
Consecutive events are segmented using onset detection, while a fundamental
frequency extraction algorithm determines their pitch.
When started with an input source (-i/--input), the detected notes are
printed on standard output, in seconds and midi note number.
When started without an input source, or with the jack option (-j/--jack),
aubionotes starts in jack mode.
OPTIONS
This program follows the usual GNU command line syntax, with long options
starting with two dashes (--). A summary of options is included below.
-i, --input source Run analysis on this audio file. Most uncompressed and
compressed are supported, depending on how aubio was built.
-r, --samplerate rate Fetch the input source, resampled at the given
sampling rate. The rate should be specified in Hertz as an integer. If 0,
the sampling rate of the original source will be used. Defaults to 0.
-B, --bufsize win The size of the buffer to analyze, that is the length
of the window used for spectral and temporal computations. Defaults to 512.
-H, --hopsize hop The number of samples between two consecutive analysis.
Defaults to 256.
-O, --onset method The onset detection method to use. See ONSET METHODS
below. Defaults to 'default'.
-t, --onset-threshold thres Set the threshold value for the onset peak
picking. Typical values are typically within 0.001 and 0.900. Defaults to
0.1. Lower threshold values imply more onsets detected. Try 0.5 in case of
over-detections. Defaults to 0.3.
-M, --minioi value Set the minimum inter-onset interval, in seconds, the
shortest interval between two consecutive notes. Defaults to 0.030
-p, --pitch method The pitch detection method to use. See PITCH METHODS
below. Defaults to 'default'.
-u, --pitch-unit unit The unit to be used to print frequencies. Possible
values include midi, bin, cent, and Hz. Defaults to 'Hz'.
-l, --pitch-tolerance thres Set the tolerance for the pitch detection
algorithm. Typical values range between 0.2 and 0.9. Pitch candidates found
with a confidence less than this threshold will not be selected. The higher
the threshold, the more confidence in the candidates. Defaults to unset.
-s, --silence sil Set the silence threshold, in dB, under which the pitch
will not be detected. A value of -20.0 would eliminate most onsets but the
loudest ones. A value of -90.0 would select all onsets. Defaults to -90.0.
-T, --timeformat format Set time format (samples, ms, seconds). Defaults to
seconds.
-j, --jack Use Jack input/output. You will need a Jack connection
controller to feed aubio some signal and listen to its output.
-h, --help Print a short help message and exit.
-v, --verbose Be verbose.
ONSET METHODS
Available methods: default, energy, hfc, complex, phase, specdiff, kl, mkl,
specflux.
See aubioonset(1) for details about these methods.
PITCH METHODS
Available methods: default, schmitt, fcomb, mcomb, specacf, yin, yinfft.
See aubiopitch(1) for details about these methods.
SEE ALSO
aubioonset(1),
aubiopitch(1),
aubiotrack(1),
aubioquiet(1),
aubiomfcc(1),
and
aubiocut(1).
AUTHOR
This manual page was written by Paul Brossier <piem@aubio.org>. Permission is
granted to copy, distribute and/or modify this document under the terms of
the GNU General Public License as published by the Free Software Foundation,
either version 3 of the License, or (at your option) any later version.
aubioquiet¶
NAME
aubioquiet - a command line tool to extracts quiet and loud regions from a file
SYNOPSIS
aubioquiet source
aubioquiet [[-i] source]
[-r rate] [-B win] [-H hop]
[-T time-format]
[-s sil]
[-v] [-h]
DESCRIPTION
aubioquiet will print a timestamp each time it detects a new silent region or
a new loud region in a sound file.
When started with an input source (-i/--input), the detected timestamps are
printed on the console, in seconds.
OPTIONS
This program follows the usual GNU command line syntax, with long options
starting with two dashes (--). A summary of options is included below.
-i, --input source Run analysis on this audio file. Most uncompressed and
compressed are supported, depending on how aubio was built.
-r, --samplerate rate Fetch the input source, resampled at the given
sampling rate. The rate should be specified in Hertz as an integer. If 0,
the sampling rate of the original source will be used. Defaults to 0.
-B, --bufsize win The size of the buffer to analyze, that is the length
of the window used for spectral and temporal computations. Defaults to 512.
-H, --hopsize hop The number of samples between two consecutive analysis.
Defaults to 256.
-s, --silence sil Set the silence threshold, in dB, under which the pitch
will not be detected. Defaults to -90.0.
-T, --timeformat format Set time format (samples, ms, seconds). Defaults to
seconds.
-h, --help Print a short help message and exit.
-v, --verbose Be verbose.
EXAMPLE OUTPUT
NOISY: 28.775330
QUIET: 28.914648
SEE ALSO
aubioonset(1),
aubiopitch(1),
aubiotrack(1),
aubionotes(1),
aubiomfcc(1),
and
aubiocut(1).
AUTHOR
This manual page was written by Paul Brossier <piem@aubio.org>. Permission
is granted to copy, distribute and/or modify this document under the terms
of the GNU General Public License, Version 3 any later version published
by the Free Software Foundation.
aubiocut¶
NAME
aubiocut - a command line tool to slice sound files at onset or beat timestamps
SYNOPSIS
aubiocut source
aubiocut [[-i] source]
[-r rate] [-B win] [-H hop]
[-O method] [-t thres]
[-b] [-c]
[-v] [-q] [-h]
OPTIONS
This program follows the usual GNU command line syntax, with long options
starting with two dashes (--). A summary of options is included below.
-i, --input source Run analysis on this audio file. Most uncompressed and
compressed are supported, depending on how aubio was built.
-r, --samplerate rate Fetch the input source, resampled at the given
sampling rate. The rate should be specified in Hertz as an integer. If set
to 0, the sampling rate of the original source will be used. Defaults to 0.
-B, --bufsize win The size of the buffer to analyze, that is the length
of the window used for spectral and temporal computations. Defaults to 512.
-H, --hopsize hop The number of samples between two consecutive analysis.
Defaults to 256.
-O, --onset method The onset detection method to use. See ONSET METHODS
below. Defaults to 'default'.
-b, --beat Use beat locations instead of onset locations.
-t, --onset-threshold thres Set the threshold value for the onset peak
picking. Values are typically in the range [0.001, 0.900]. Lower threshold
values imply more onsets detected. Increasing this threshold should reduce
the number of incorrect detections. Defaults to 0.3.
-c, --cut Cut input sound file at detected labels. A new sound files for
each slice will be created in the current directory.
-o, --output directory Specify the directory path where slices of the
original source should be created.
--cut-until-nsamples n How many extra samples should be added at the end of
each slice (default 0).
--cut-until-nslices n How many extra slices should be added at the end of
each slice (default 0).
-h, --help Print a short help message and exit.
-v, --verbose Be verbose.
-q, --quiet Be quiet.
ONSET METHODS
Available methods: default, energy, hfc, complex, phase, specdiff, kl, mkl,
specflux.
See aubioonset(1) for details about these methods.
SEE ALSO
aubioonset(1),
aubiopitch(1),
aubiotrack(1),
aubionotes(1),
aubioquiet(1),
and
aubiomfcc(1).
AUTHOR
This manual page was written by Paul Brossier <piem@aubio.org>. Permission is
granted to copy, distribute and/or modify this document under the terms of
the GNU General Public License as published by the Free Software Foundation,
either version 3 of the License, or (at your option) any later version.