ref: 8e2f36a607e19683241bdd45894905346c9f9dea
parent: 1f00b698e39f0985bc575284d9e614d3e8bcedbc
author: Paul Brossier <piem@piem.org>
date: Thu Mar 23 10:56:50 EDT 2017
setup.py: move cut and cmd inside aubio module, use entry_points to generate scripts
--- /dev/null
+++ b/python/lib/aubio/cmd.py
@@ -1,0 +1,370 @@
+#! /usr/bin/env python
+# -*- coding: utf-8 -*-
+
+"""aubio command line tool
+
+This file was written by Paul Brossier <piem@aubio.org> and is released under
+the GNU/GPL v3.
+
+Note: this script is mostly about parsing command line arguments. For more
+readable code examples, check out the `python/demos` folder."""
+
+import sys
+import argparse
+import aubio
+
+def aubio_parser():
+ epilog = 'use "%(prog)s <command> --help" for more info about each command'
+ parser = argparse.ArgumentParser(epilog=epilog)
+ parser.add_argument('-V', '--version', help="show version",
+ action="store_true", dest="show_version")
+
+ subparsers = parser.add_subparsers(dest='command',
+ description="", metavar="<command>")
+
+ # onset subcommand
+ subparser = subparsers.add_parser('onset',
+ help='get onset times',
+ formatter_class = argparse.ArgumentDefaultsHelpFormatter)
+ parser_add_input(subparser)
+ parser_add_buf_hop_size(subparser)
+ helpstr = "onset novelty function"
+ helpstr += " <default|energy|hfc|complex|phase|specdiff|kl|mkl|specflux>"
+ parser_add_method(subparser, helpstr=helpstr)
+ parser_add_threshold(subparser)
+ parser_add_silence(subparser)
+ parser_add_minioi(subparser)
+ parser_add_time_format(subparser)
+ parser_add_verbose_help(subparser)
+ subparser.set_defaults(process=process_onset)
+
+ # pitch subcommand
+ subparser = subparsers.add_parser('pitch',
+ help='extract fundamental frequency')
+ parser_add_input(subparser)
+ parser_add_buf_hop_size(subparser, buf_size=2048)
+ helpstr = "pitch detection method <default|yinfft|yin|mcomb|fcomb|schmitt>"
+ parser_add_method(subparser, helpstr=helpstr)
+ parser_add_threshold(subparser)
+ parser_add_silence(subparser)
+ parser_add_time_format(subparser)
+ parser_add_verbose_help(subparser)
+ subparser.set_defaults(process=process_pitch)
+
+ # tempo subcommand
+ subparser = subparsers.add_parser('beat',
+ help='get locations of beats')
+ parser_add_input(subparser)
+ parser_add_buf_hop_size(subparser, buf_size=1024, hop_size=512)
+ parser_add_time_format(subparser)
+ parser_add_verbose_help(subparser)
+ subparser.set_defaults(process=process_tempo)
+
+ # notes subcommand
+ subparser = subparsers.add_parser('notes',
+ help='get midi-like notes')
+ parser_add_input(subparser)
+ parser_add_buf_hop_size(subparser)
+ parser_add_time_format(subparser)
+ parser_add_verbose_help(subparser)
+ subparser.set_defaults(process=process_notes)
+
+ # mfcc subcommand
+ subparser = subparsers.add_parser('mfcc',
+ help='extract mel-frequency cepstrum coefficients')
+ parser_add_input(subparser)
+ parser_add_buf_hop_size(subparser)
+ parser_add_time_format(subparser)
+ parser_add_verbose_help(subparser)
+ subparser.set_defaults(process=process_mfcc)
+
+ # melbands subcommand
+ subparser = subparsers.add_parser('melbands',
+ help='extract mel-frequency energies per band')
+ parser_add_input(subparser)
+ parser_add_buf_hop_size(subparser)
+ parser_add_time_format(subparser)
+ parser_add_verbose_help(subparser)
+ subparser.set_defaults(process=process_melbands)
+
+ return parser
+
+def parser_add_input(parser):
+ parser.add_argument("source_uri", default=None, nargs='?',
+ help="input sound file to analyse", metavar = "<source_uri>")
+ parser.add_argument("-i", "--input", dest = "source_uri2",
+ help="input sound file to analyse", metavar = "<source_uri>")
+ parser.add_argument("-r", "--samplerate",
+ metavar = "<freq>", type=int,
+ action="store", dest="samplerate", default=0,
+ help="samplerate at which the file should be represented")
+
+def parser_add_verbose_help(parser):
+ parser.add_argument("-v","--verbose",
+ action="count", dest="verbose", default=1,
+ help="make lots of noise [default]")
+ parser.add_argument("-q","--quiet",
+ action="store_const", dest="verbose", const=0,
+ help="be quiet")
+
+def parser_add_buf_hop_size(parser, buf_size=512, hop_size=256):
+ parser.add_argument("-B","--bufsize",
+ action="store", dest="buf_size", default=buf_size,
+ metavar = "<size>", type=int,
+ help="buffer size [default=%d]" % buf_size)
+ parser.add_argument("-H","--hopsize",
+ metavar = "<size>", type=int,
+ action="store", dest="hop_size", default=hop_size,
+ help="overlap size [default=%d]" % hop_size)
+
+def parser_add_method(parser, method='default', helpstr='method'):
+ parser.add_argument("-m","--method",
+ metavar = "<method>", type=str,
+ action="store", dest="method", default=method,
+ help="%s [default=%s]" % (helpstr, method))
+
+def parser_add_threshold(parser, default=None):
+ parser.add_argument("-t","--threshold",
+ metavar = "<threshold>", type=float,
+ action="store", dest="threshold", default=default,
+ help="threshold [default=%s]" % default)
+
+def parser_add_silence(parser):
+ parser.add_argument("-s", "--silence",
+ metavar = "<value>", type=float,
+ action="store", dest="silence", default=-70,
+ help="silence threshold")
+
+def parser_add_minioi(parser):
+ parser.add_argument("-M", "--minioi",
+ metavar = "<value>", type=str,
+ action="store", dest="minioi", default="12ms",
+ help="minimum Inter-Onset Interval")
+
+def parser_add_time_format(parser):
+ helpstr = "select time values output format (samples, ms, seconds)"
+ helpstr += " [default=seconds]"
+ parser.add_argument("-T", "--time-format",
+ metavar='format',
+ dest="time_format",
+ default=None,
+ help=helpstr)
+
+# some utilities
+
+def parse_options(args, valid_opts):
+ options = {k :v for k,v in vars(args).items() if k in valid_opts}
+ return options
+
+def remap_pvoc_options(options):
+ # remap buf_size to win_s, hop_size to hop_s
+ # FIXME: adjust python/ext/py-phasevoc.c to understand buf_size/hop_size
+ options['win_s'] = options['buf_size']
+ del options['buf_size']
+ options['hop_s'] = options['hop_size']
+ del options['hop_size']
+ return options
+
+def samples2seconds(n_frames, samplerate):
+ return "%f\t" % (n_frames / float(a_source.samplerate))
+
+def samples2milliseconds(n_frames, samplerate):
+ return "%f\t" % (1000. * n_frames / float(a_source.samplerate))
+
+def samples2samples(n_frames, samplerate):
+ return "%d\t" % n_frames
+
+def timefunc(mode):
+ if mode is None or mode == 'seconds' or mode == 's':
+ return samples2seconds
+ elif mode == 'ms' or mode == 'milliseconds':
+ return samples2milliseconds
+ elif mode == 'samples':
+ return samples2samples
+ else:
+ raise ValueError('invalid time format %s' % mode)
+
+# definition of processing classes
+
+class default_process(object):
+ def __init__(self, args):
+ if 'time_format' in args:
+ self.time2string = timefunc(args.time_format)
+ if args.verbose > 2 and hasattr(self, 'options'):
+ name = type(self).__name__.split('_')[1]
+ optstr = ' '.join(['running', name, 'with options', repr(self.options), '\n'])
+ sys.stderr.write(optstr)
+
+class process_onset(default_process):
+ valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
+ def __init__(self, args):
+ self.options = parse_options(args, self.valid_opts)
+ self.onset = aubio.onset(**self.options)
+ if args.threshold is not None:
+ self.onset.set_threshold(args.threshold)
+ if args.minioi:
+ if args.minioi.endswith('ms'):
+ self.onset.set_minioi_ms(float(args.minioi[:-2]))
+ elif args.minioi.endswith('s'):
+ self.onset.set_minioi_s(float(args.minioi[:-1]))
+ else:
+ self.onset.set_minioi(int(args.minioi))
+ if args.silence:
+ self.onset.set_silence(args.silence)
+ super(process_onset, self).__init__(args)
+ def __call__(self, block):
+ return self.onset(block)
+ def repr_res(self, res, frames_read, a_source):
+ if res[0] != 0:
+ outstr = self.time2string(self.onset.get_last(), a_source.samplerate)
+ sys.stdout.write(outstr + '\n')
+
+class process_pitch(default_process):
+ valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
+ def __init__(self, args):
+ self.options = parse_options(args, self.valid_opts)
+ self.pitch = aubio.pitch(**self.options)
+ if args.threshold is not None:
+ self.pitch.set_tolerance(args.threshold)
+ if args.silence is not None:
+ self.pitch.set_silence(args.silence)
+ super(process_pitch, self).__init__(args)
+ def __call__(self, block):
+ return self.pitch(block)
+ def repr_res(self, res, frames_read, a_source):
+ fmt_out = self.time2string(frames_read, a_source.samplerate)
+ sys.stdout.write(fmt_out + "%.6f\n" % res[0])
+
+class process_tempo(default_process):
+ valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
+ def __init__(self, args):
+ self.options = parse_options(args, self.valid_opts)
+ self.tempo = aubio.tempo(**self.options)
+ super(process_tempo, self).__init__(args)
+ def __call__(self, block):
+ return self.tempo(block)
+ def repr_res(self, res, frames_read, a_source):
+ if res[0] != 0:
+ outstr = self.time2string(self.tempo.get_last(), a_source.samplerate)
+ sys.stdout.write(outstr + '\n')
+
+class process_notes(default_process):
+ valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
+ def __init__(self, args):
+ self.options = parse_options(args, self.valid_opts)
+ self.notes = aubio.notes(**self.options)
+ super(process_notes, self).__init__(args)
+ def __call__(self, block):
+ return self.notes(block)
+ def repr_res(self, res, frames_read, a_source):
+ if res[2] != 0: # note off
+ fmt_out = self.time2string(frames_read, a_source.samplerate)
+ sys.stdout.write(fmt_out + '\n')
+ if res[0] != 0: # note on
+ lastmidi = res[0]
+ fmt_out = "%f\t" % lastmidi
+ fmt_out += self.time2string(frames_read, a_source.samplerate)
+ sys.stdout.write(fmt_out) # + '\t')
+
+class process_mfcc(default_process):
+ def __init__(self, args):
+ valid_opts = ['hop_size', 'buf_size']
+ options = parse_options(args, valid_opts)
+ self.options = remap_pvoc_options(options)
+ self.pv = aubio.pvoc(**options)
+
+ valid_opts = ['buf_size', 'n_filters', 'n_coeffs', 'samplerate']
+ options = parse_options(args, valid_opts)
+ self.mfcc = aubio.mfcc(**options)
+ self.options.update(options)
+
+ super(process_mfcc, self).__init__(args)
+
+ def __call__(self, block):
+ fftgrain = self.pv(block)
+ return self.mfcc(fftgrain)
+ def repr_res(self, res, frames_read, a_source):
+ fmt_out = self.time2string(frames_read, a_source.samplerate)
+ fmt_out += ' '.join(["% 9.7f" % f for f in res.tolist()])
+ sys.stdout.write(fmt_out + '\n')
+
+class process_melbands(default_process):
+ def __init__(self, args):
+ self.args = args
+ valid_opts = ['hop_size', 'buf_size']
+ options = parse_options(args, valid_opts)
+ options = remap_pvoc_options(options)
+ self.pv = aubio.pvoc(**options)
+
+ valid_opts = ['buf_size', 'n_filters']
+ options = {k :v for k,v in vars(args).items() if k in valid_opts}
+ # FIXME
+ options['win_s'] = options['buf_size']
+ del options['buf_size']
+ self.filterbank = aubio.filterbank(**options)
+ self.filterbank.set_mel_coeffs_slaney(args.samplerate)
+
+ super(process_melbands, self).__init__(args)
+ def __call__(self, block):
+ fftgrain = self.pv(block)
+ return self.filterbank(fftgrain)
+ def repr_res(self, res, frames_read, a_source):
+ fmt_out = self.time2string(frames_read, a_source.samplerate)
+ fmt_out += ' '.join(["% 9.7f" % f for f in res.tolist()])
+ sys.stdout.write(fmt_out + '\n')
+
+def main():
+ parser = aubio_parser()
+ args = parser.parse_args()
+ if args.show_version or ('verbose' in args and args.verbose > 3):
+ sys.stdout.write('aubio version ' + aubio.version + '\n')
+ if args.show_version and args.command is None:
+ sys.exit(0)
+ if args.command is None:
+ sys.stderr.write("Error: a command is required\n")
+ parser.print_help()
+ sys.exit(1)
+ elif not args.source_uri and not args.source_uri2:
+ sys.stderr.write("Error: a source is required\n")
+ parser.print_help()
+ sys.exit(1)
+ elif args.source_uri2 is not None:
+ args.source_uri = args.source_uri2
+ try:
+ # open source_uri
+ with aubio.source(args.source_uri, hop_size=args.hop_size,
+ samplerate=args.samplerate) as a_source:
+ args.samplerate = a_source.samplerate
+ # create the processor for this subcommand
+ processor = args.process(args)
+ frames_read = 0
+ while True:
+ # read new block from source
+ block, read = a_source()
+ # execute processor on this block
+ res = processor(block)
+ # print results for this block
+ if args.verbose > 0:
+ processor.repr_res(res, frames_read, a_source)
+ # increment total number of frames read
+ frames_read += read
+ # exit loop at end of file
+ if read < a_source.hop_size: break
+ # special case for notes
+ if args.command == 'notes':
+ eof = processor.time2string(frames_read, a_source.samplerate)
+ sys.stdout.write(eof + '\n')
+ sys.stdout.flush()
+ if args.verbose > 1:
+ fmt_string = "read {:.2f}s"
+ fmt_string += " ({:d} samples in {:d} blocks of {:d})"
+ fmt_string += " from {:s} at {:d}Hz\n"
+ sys.stderr.write(fmt_string.format(
+ frames_read/float(a_source.samplerate),
+ frames_read,
+ frames_read // a_source.hop_size + 1,
+ a_source.hop_size,
+ a_source.uri,
+ a_source.samplerate))
+ except KeyboardInterrupt as e:
+ sys.exit(1)
--- /dev/null
+++ b/python/lib/aubio/cut.py
@@ -1,0 +1,211 @@
+#! /usr/bin/env python
+
+""" this file was written by Paul Brossier
+ it is released under the GNU/GPL license.
+"""
+
+import sys
+
+def parse_args():
+ from optparse import OptionParser
+ usage = "usage: %s [options] -i soundfile" % sys.argv[0]
+ usage += "\n help: %s -h" % sys.argv[0]
+ parser = OptionParser(usage=usage)
+ parser.add_option("-i", "--input", action = "store", dest = "source_file",
+ help="input sound file to analyse", metavar = "<source_file>")
+ parser.add_option("-O","--onset-method",
+ action="store", dest="onset_method", default='default',
+ metavar = "<onset_method>",
+ help="onset detection method [default=default] \
+ complexdomain|hfc|phase|specdiff|energy|kl|mkl")
+ # cutting methods
+ parser.add_option("-b","--beat",
+ action="store_true", dest="beat", default=False,
+ help="use beat locations")
+ """
+ parser.add_option("-S","--silencecut",
+ action="store_true", dest="silencecut", default=False,
+ help="use silence locations")
+ parser.add_option("-s","--silence",
+ metavar = "<value>",
+ action="store", dest="silence", default=-70,
+ help="silence threshold [default=-70]")
+ """
+ # algorithm parameters
+ parser.add_option("-r", "--samplerate",
+ metavar = "<freq>", type='int',
+ action="store", dest="samplerate", default=0,
+ help="samplerate at which the file should be represented")
+ parser.add_option("-B","--bufsize",
+ action="store", dest="bufsize", default=512,
+ metavar = "<size>", type='int',
+ help="buffer size [default=512]")
+ parser.add_option("-H","--hopsize",
+ metavar = "<size>", type='int',
+ action="store", dest="hopsize", default=256,
+ help="overlap size [default=256]")
+ parser.add_option("-t","--onset-threshold",
+ metavar = "<value>", type="float",
+ action="store", dest="threshold", default=0.3,
+ help="onset peak picking threshold [default=0.3]")
+ parser.add_option("-c","--cut",
+ action="store_true", dest="cut", default=False,
+ help="cut input sound file at detected labels \
+ best used with option -L")
+
+ # minioi
+ parser.add_option("-M","--minioi",
+ metavar = "<value>", type='string',
+ action="store", dest="minioi", default="12ms",
+ help="minimum inter onset interval [default=12ms]")
+
+ """
+ parser.add_option("-D","--delay",
+ action = "store", dest = "delay", type = "float",
+ metavar = "<seconds>", default=0,
+ help="number of seconds to take back [default=system]\
+ default system delay is 3*hopsize/samplerate")
+ parser.add_option("-C","--dcthreshold",
+ metavar = "<value>",
+ action="store", dest="dcthreshold", default=1.,
+ help="onset peak picking DC component [default=1.]")
+ parser.add_option("-L","--localmin",
+ action="store_true", dest="localmin", default=False,
+ help="use local minima after peak detection")
+ parser.add_option("-d","--derivate",
+ action="store_true", dest="derivate", default=False,
+ help="derivate onset detection function")
+ parser.add_option("-z","--zerocross",
+ metavar = "<value>",
+ action="store", dest="zerothres", default=0.008,
+ help="zero-crossing threshold for slicing [default=0.00008]")
+ """
+ # plotting functions
+ """
+ parser.add_option("-p","--plot",
+ action="store_true", dest="plot", default=False,
+ help="draw plot")
+ parser.add_option("-x","--xsize",
+ metavar = "<size>",
+ action="store", dest="xsize", default=1.,
+ type='float', help="define xsize for plot")
+ parser.add_option("-y","--ysize",
+ metavar = "<size>",
+ action="store", dest="ysize", default=1.,
+ type='float', help="define ysize for plot")
+ parser.add_option("-f","--function",
+ action="store_true", dest="func", default=False,
+ help="print detection function")
+ parser.add_option("-n","--no-onsets",
+ action="store_true", dest="nplot", default=False,
+ help="do not plot detected onsets")
+ parser.add_option("-O","--outplot",
+ metavar = "<output_image>",
+ action="store", dest="outplot", default=None,
+ help="save plot to output.{ps,png}")
+ parser.add_option("-F","--spectrogram",
+ action="store_true", dest="spectro", default=False,
+ help="add spectrogram to the plot")
+ """
+ parser.add_option("-o","--output", type = str,
+ metavar = "<outputdir>",
+ action="store", dest="output_directory", default=None,
+ help="specify path where slices of the original file should be created")
+ parser.add_option("--cut-until-nsamples", type = int,
+ metavar = "<samples>",
+ action = "store", dest = "cut_until_nsamples", default = None,
+ help="how many extra samples should be added at the end of each slice")
+ parser.add_option("--cut-every-nslices", type = int,
+ metavar = "<samples>",
+ action = "store", dest = "cut_every_nslices", default = None,
+ help="how many slices should be groupped together at each cut")
+ parser.add_option("--cut-until-nslices", type = int,
+ metavar = "<slices>",
+ action = "store", dest = "cut_until_nslices", default = None,
+ help="how many extra slices should be added at the end of each slice")
+
+ parser.add_option("-v","--verbose",
+ action="store_true", dest="verbose", default=True,
+ help="make lots of noise [default]")
+ parser.add_option("-q","--quiet",
+ action="store_false", dest="verbose", default=True,
+ help="be quiet")
+ (options, args) = parser.parse_args()
+ if not options.source_file:
+ import os.path
+ if len(args) == 1:
+ options.source_file = args[0]
+ else:
+ print ("no file name given\n" + usage)
+ sys.exit(1)
+ return options, args
+
+def main():
+ options, args = parse_args()
+
+ hopsize = options.hopsize
+ bufsize = options.bufsize
+ samplerate = options.samplerate
+ source_file = options.source_file
+
+ from aubio import onset, tempo, source, sink
+
+ s = source(source_file, samplerate, hopsize)
+ if samplerate == 0: samplerate = s.get_samplerate()
+
+ if options.beat:
+ o = tempo(options.onset_method, bufsize, hopsize)
+ else:
+ o = onset(options.onset_method, bufsize, hopsize)
+ if options.minioi:
+ if options.minioi.endswith('ms'):
+ o.set_minioi_ms(int(options.minioi[:-2]))
+ elif options.minioi.endswith('s'):
+ o.set_minioi_s(int(options.minioi[:-1]))
+ else:
+ o.set_minioi(int(options.minioi))
+ o.set_threshold(options.threshold)
+
+ timestamps = []
+ total_frames = 0
+ # analyze pass
+ while True:
+ samples, read = s()
+ if o(samples):
+ timestamps.append (o.get_last())
+ if options.verbose: print ("%.4f" % o.get_last_s())
+ total_frames += read
+ if read < hopsize: break
+ del s
+ # print some info
+ nstamps = len(timestamps)
+ duration = float (total_frames) / float(samplerate)
+ info = 'found %(nstamps)d timestamps in %(source_file)s' % locals()
+ info += ' (total %(duration).2fs at %(samplerate)dHz)\n' % locals()
+ sys.stderr.write(info)
+
+ # cutting pass
+ if options.cut and nstamps > 0:
+ # generate output files
+ from aubio.slicing import slice_source_at_stamps
+ timestamps_end = None
+ if options.cut_every_nslices:
+ timestamps = timestamps[::options.cut_every_nslices]
+ nstamps = len(timestamps)
+ if options.cut_until_nslices and options.cut_until_nsamples:
+ print ("warning: using cut_until_nslices, but cut_until_nsamples is set")
+ if options.cut_until_nsamples:
+ timestamps_end = [t + options.cut_until_nsamples for t in timestamps[1:]]
+ timestamps_end += [ 1e120 ]
+ if options.cut_until_nslices:
+ timestamps_end = [t for t in timestamps[1 + options.cut_until_nslices:]]
+ timestamps_end += [ 1e120 ] * (options.cut_until_nslices + 1)
+ slice_source_at_stamps(source_file, timestamps, timestamps_end = timestamps_end,
+ output_dir = options.output_directory,
+ samplerate = samplerate)
+
+ # print some info
+ duration = float (total_frames) / float(samplerate)
+ info = 'created %(nstamps)d slices from %(source_file)s' % locals()
+ info += ' (total %(duration).2fs at %(samplerate)dHz)\n' % locals()
+ sys.stderr.write(info)
--- a/python/scripts/aubio
+++ /dev/null
@@ -1,370 +1,0 @@
-#! /usr/bin/env python
-# -*- coding: utf-8 -*-
-
-"""aubio command line tool
-
-This file was written by Paul Brossier <piem@aubio.org> and is released under
-the GNU/GPL v3.
-
-Note: this script is mostly about parsing command line arguments. For more
-readable code examples, check out the `python/demos` folder."""
-
-import sys
-import argparse
-import aubio
-
-def aubio_parser():
- epilog = 'use "%(prog)s <command> --help" for more info about each command'
- parser = argparse.ArgumentParser(epilog=epilog)
- parser.add_argument('-V', '--version', help="show version",
- action="store_true", dest="show_version")
-
- subparsers = parser.add_subparsers(dest='command',
- description="", metavar="<command>")
-
- # onset subcommand
- subparser = subparsers.add_parser('onset',
- help='get onset times',
- formatter_class = argparse.ArgumentDefaultsHelpFormatter)
- parser_add_input(subparser)
- parser_add_buf_hop_size(subparser)
- helpstr = "onset novelty function"
- helpstr += " <default|energy|hfc|complex|phase|specdiff|kl|mkl|specflux>"
- parser_add_method(subparser, helpstr=helpstr)
- parser_add_threshold(subparser)
- parser_add_silence(subparser)
- parser_add_minioi(subparser)
- parser_add_time_format(subparser)
- parser_add_verbose_help(subparser)
- subparser.set_defaults(process=process_onset)
-
- # pitch subcommand
- subparser = subparsers.add_parser('pitch',
- help='extract fundamental frequency')
- parser_add_input(subparser)
- parser_add_buf_hop_size(subparser, buf_size=2048)
- helpstr = "pitch detection method <default|yinfft|yin|mcomb|fcomb|schmitt>"
- parser_add_method(subparser, helpstr=helpstr)
- parser_add_threshold(subparser)
- parser_add_silence(subparser)
- parser_add_time_format(subparser)
- parser_add_verbose_help(subparser)
- subparser.set_defaults(process=process_pitch)
-
- # tempo subcommand
- subparser = subparsers.add_parser('beat',
- help='get locations of beats')
- parser_add_input(subparser)
- parser_add_buf_hop_size(subparser, buf_size=1024, hop_size=512)
- parser_add_time_format(subparser)
- parser_add_verbose_help(subparser)
- subparser.set_defaults(process=process_tempo)
-
- # notes subcommand
- subparser = subparsers.add_parser('notes',
- help='get midi-like notes')
- parser_add_input(subparser)
- parser_add_buf_hop_size(subparser)
- parser_add_time_format(subparser)
- parser_add_verbose_help(subparser)
- subparser.set_defaults(process=process_notes)
-
- # mfcc subcommand
- subparser = subparsers.add_parser('mfcc',
- help='extract mel-frequency cepstrum coefficients')
- parser_add_input(subparser)
- parser_add_buf_hop_size(subparser)
- parser_add_time_format(subparser)
- parser_add_verbose_help(subparser)
- subparser.set_defaults(process=process_mfcc)
-
- # melbands subcommand
- subparser = subparsers.add_parser('melbands',
- help='extract mel-frequency energies per band')
- parser_add_input(subparser)
- parser_add_buf_hop_size(subparser)
- parser_add_time_format(subparser)
- parser_add_verbose_help(subparser)
- subparser.set_defaults(process=process_melbands)
-
- return parser
-
-def parser_add_input(parser):
- parser.add_argument("source_uri", default=None, nargs='?',
- help="input sound file to analyse", metavar = "<source_uri>")
- parser.add_argument("-i", "--input", dest = "source_uri2",
- help="input sound file to analyse", metavar = "<source_uri>")
- parser.add_argument("-r", "--samplerate",
- metavar = "<freq>", type=int,
- action="store", dest="samplerate", default=0,
- help="samplerate at which the file should be represented")
-
-def parser_add_verbose_help(parser):
- parser.add_argument("-v","--verbose",
- action="count", dest="verbose", default=1,
- help="make lots of noise [default]")
- parser.add_argument("-q","--quiet",
- action="store_const", dest="verbose", const=0,
- help="be quiet")
-
-def parser_add_buf_hop_size(parser, buf_size=512, hop_size=256):
- parser.add_argument("-B","--bufsize",
- action="store", dest="buf_size", default=buf_size,
- metavar = "<size>", type=int,
- help="buffer size [default=%d]" % buf_size)
- parser.add_argument("-H","--hopsize",
- metavar = "<size>", type=int,
- action="store", dest="hop_size", default=hop_size,
- help="overlap size [default=%d]" % hop_size)
-
-def parser_add_method(parser, method='default', helpstr='method'):
- parser.add_argument("-m","--method",
- metavar = "<method>", type=str,
- action="store", dest="method", default=method,
- help="%s [default=%s]" % (helpstr, method))
-
-def parser_add_threshold(parser, default=None):
- parser.add_argument("-t","--threshold",
- metavar = "<threshold>", type=float,
- action="store", dest="threshold", default=default,
- help="threshold [default=%s]" % default)
-
-def parser_add_silence(parser):
- parser.add_argument("-s", "--silence",
- metavar = "<value>", type=float,
- action="store", dest="silence", default=-70,
- help="silence threshold")
-
-def parser_add_minioi(parser):
- parser.add_argument("-M", "--minioi",
- metavar = "<value>", type=str,
- action="store", dest="minioi", default="12ms",
- help="minimum Inter-Onset Interval")
-
-def parser_add_time_format(parser):
- helpstr = "select time values output format (samples, ms, seconds)"
- helpstr += " [default=seconds]"
- parser.add_argument("-T", "--time-format",
- metavar='format',
- dest="time_format",
- default=None,
- help=helpstr)
-
-# some utilities
-
-def parse_options(args, valid_opts):
- options = {k :v for k,v in vars(args).items() if k in valid_opts}
- return options
-
-def remap_pvoc_options(options):
- # remap buf_size to win_s, hop_size to hop_s
- # FIXME: adjust python/ext/py-phasevoc.c to understand buf_size/hop_size
- options['win_s'] = options['buf_size']
- del options['buf_size']
- options['hop_s'] = options['hop_size']
- del options['hop_size']
- return options
-
-def samples2seconds(n_frames, samplerate):
- return "%f\t" % (n_frames / float(a_source.samplerate))
-
-def samples2milliseconds(n_frames, samplerate):
- return "%f\t" % (1000. * n_frames / float(a_source.samplerate))
-
-def samples2samples(n_frames, samplerate):
- return "%d\t" % n_frames
-
-def timefunc(mode):
- if mode is None or mode == 'seconds' or mode == 's':
- return samples2seconds
- elif mode == 'ms' or mode == 'milliseconds':
- return samples2milliseconds
- elif mode == 'samples':
- return samples2samples
- else:
- raise ValueError('invalid time format %s' % mode)
-
-# definition of processing classes
-
-class default_process(object):
- def __init__(self, args):
- if 'time_format' in args:
- self.time2string = timefunc(args.time_format)
- if args.verbose > 2 and hasattr(self, 'options'):
- name = type(self).__name__.split('_')[1]
- optstr = ' '.join(['running', name, 'with options', repr(self.options), '\n'])
- sys.stderr.write(optstr)
-
-class process_onset(default_process):
- valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
- def __init__(self, args):
- self.options = parse_options(args, self.valid_opts)
- self.onset = aubio.onset(**self.options)
- if args.threshold is not None:
- self.onset.set_threshold(args.threshold)
- if args.minioi:
- if args.minioi.endswith('ms'):
- self.onset.set_minioi_ms(float(args.minioi[:-2]))
- elif args.minioi.endswith('s'):
- self.onset.set_minioi_s(float(args.minioi[:-1]))
- else:
- self.onset.set_minioi(int(args.minioi))
- if args.silence:
- self.onset.set_silence(args.silence)
- super(process_onset, self).__init__(args)
- def __call__(self, block):
- return self.onset(block)
- def repr_res(self, res, frames_read, a_source):
- if res[0] != 0:
- outstr = self.time2string(self.onset.get_last(), a_source.samplerate)
- sys.stdout.write(outstr + '\n')
-
-class process_pitch(default_process):
- valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
- def __init__(self, args):
- self.options = parse_options(args, self.valid_opts)
- self.pitch = aubio.pitch(**self.options)
- if args.threshold is not None:
- self.pitch.set_tolerance(args.threshold)
- if args.silence is not None:
- self.pitch.set_silence(args.silence)
- super(process_pitch, self).__init__(args)
- def __call__(self, block):
- return self.pitch(block)
- def repr_res(self, res, frames_read, a_source):
- fmt_out = self.time2string(frames_read, a_source.samplerate)
- sys.stdout.write(fmt_out + "%.6f\n" % res[0])
-
-class process_tempo(default_process):
- valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
- def __init__(self, args):
- self.options = parse_options(args, self.valid_opts)
- self.tempo = aubio.tempo(**self.options)
- super(process_tempo, self).__init__(args)
- def __call__(self, block):
- return self.tempo(block)
- def repr_res(self, res, frames_read, a_source):
- if res[0] != 0:
- outstr = self.time2string(self.tempo.get_last(), a_source.samplerate)
- sys.stdout.write(outstr + '\n')
-
-class process_notes(default_process):
- valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
- def __init__(self, args):
- self.options = parse_options(args, self.valid_opts)
- self.notes = aubio.notes(**self.options)
- super(process_notes, self).__init__(args)
- def __call__(self, block):
- return self.notes(block)
- def repr_res(self, res, frames_read, a_source):
- if res[2] != 0: # note off
- fmt_out = self.time2string(frames_read, a_source.samplerate)
- sys.stdout.write(fmt_out + '\n')
- if res[0] != 0: # note on
- lastmidi = res[0]
- fmt_out = "%f\t" % lastmidi
- fmt_out += self.time2string(frames_read, a_source.samplerate)
- sys.stdout.write(fmt_out) # + '\t')
-
-class process_mfcc(default_process):
- def __init__(self, args):
- valid_opts = ['hop_size', 'buf_size']
- options = parse_options(args, valid_opts)
- self.options = remap_pvoc_options(options)
- self.pv = aubio.pvoc(**options)
-
- valid_opts = ['buf_size', 'n_filters', 'n_coeffs', 'samplerate']
- options = parse_options(args, valid_opts)
- self.mfcc = aubio.mfcc(**options)
- self.options.update(options)
-
- super(process_mfcc, self).__init__(args)
-
- def __call__(self, block):
- fftgrain = self.pv(block)
- return self.mfcc(fftgrain)
- def repr_res(self, res, frames_read, a_source):
- fmt_out = self.time2string(frames_read, a_source.samplerate)
- fmt_out += ' '.join(["% 9.7f" % f for f in res.tolist()])
- sys.stdout.write(fmt_out + '\n')
-
-class process_melbands(default_process):
- def __init__(self, args):
- self.args = args
- valid_opts = ['hop_size', 'buf_size']
- options = parse_options(args, valid_opts)
- options = remap_pvoc_options(options)
- self.pv = aubio.pvoc(**options)
-
- valid_opts = ['buf_size', 'n_filters']
- options = {k :v for k,v in vars(args).items() if k in valid_opts}
- # FIXME
- options['win_s'] = options['buf_size']
- del options['buf_size']
- self.filterbank = aubio.filterbank(**options)
- self.filterbank.set_mel_coeffs_slaney(args.samplerate)
-
- super(process_melbands, self).__init__(args)
- def __call__(self, block):
- fftgrain = self.pv(block)
- return self.filterbank(fftgrain)
- def repr_res(self, res, frames_read, a_source):
- fmt_out = self.time2string(frames_read, a_source.samplerate)
- fmt_out += ' '.join(["% 9.7f" % f for f in res.tolist()])
- sys.stdout.write(fmt_out + '\n')
-
-if __name__ == '__main__':
- parser = aubio_parser()
- args = parser.parse_args()
- if args.show_version or ('verbose' in args and args.verbose > 3):
- sys.stdout.write('aubio version ' + aubio.version + '\n')
- if args.show_version and args.command is None:
- sys.exit(0)
- if args.command is None:
- sys.stderr.write("Error: a command is required\n")
- parser.print_help()
- sys.exit(1)
- elif not args.source_uri and not args.source_uri2:
- sys.stderr.write("Error: a source is required\n")
- parser.print_help()
- sys.exit(1)
- elif args.source_uri2 is not None:
- args.source_uri = args.source_uri2
- try:
- # open source_uri
- with aubio.source(args.source_uri, hop_size=args.hop_size,
- samplerate=args.samplerate) as a_source:
- args.samplerate = a_source.samplerate
- # create the processor for this subcommand
- processor = args.process(args)
- frames_read = 0
- while True:
- # read new block from source
- block, read = a_source()
- # execute processor on this block
- res = processor(block)
- # print results for this block
- if args.verbose > 0:
- processor.repr_res(res, frames_read, a_source)
- # increment total number of frames read
- frames_read += read
- # exit loop at end of file
- if read < a_source.hop_size: break
- # special case for notes
- if args.command == 'notes':
- eof = processor.time2string(frames_read, a_source.samplerate)
- sys.stdout.write(eof + '\n')
- sys.stdout.flush()
- if args.verbose > 1:
- fmt_string = "read {:.2f}s"
- fmt_string += " ({:d} samples in {:d} blocks of {:d})"
- fmt_string += " from {:s} at {:d}Hz\n"
- sys.stderr.write(fmt_string.format(
- frames_read/float(a_source.samplerate),
- frames_read,
- frames_read // a_source.hop_size + 1,
- a_source.hop_size,
- a_source.uri,
- a_source.samplerate))
- except KeyboardInterrupt as e:
- sys.exit(1)
--- a/python/scripts/aubiocut
+++ /dev/null
@@ -1,211 +1,0 @@
-#! /usr/bin/env python
-
-""" this file was written by Paul Brossier
- it is released under the GNU/GPL license.
-"""
-
-import sys
-
-def parse_args():
- from optparse import OptionParser
- usage = "usage: %s [options] -i soundfile" % sys.argv[0]
- usage += "\n help: %s -h" % sys.argv[0]
- parser = OptionParser(usage=usage)
- parser.add_option("-i", "--input", action = "store", dest = "source_file",
- help="input sound file to analyse", metavar = "<source_file>")
- parser.add_option("-O","--onset-method",
- action="store", dest="onset_method", default='default',
- metavar = "<onset_method>",
- help="onset detection method [default=default] \
- complexdomain|hfc|phase|specdiff|energy|kl|mkl")
- # cutting methods
- parser.add_option("-b","--beat",
- action="store_true", dest="beat", default=False,
- help="use beat locations")
- """
- parser.add_option("-S","--silencecut",
- action="store_true", dest="silencecut", default=False,
- help="use silence locations")
- parser.add_option("-s","--silence",
- metavar = "<value>",
- action="store", dest="silence", default=-70,
- help="silence threshold [default=-70]")
- """
- # algorithm parameters
- parser.add_option("-r", "--samplerate",
- metavar = "<freq>", type='int',
- action="store", dest="samplerate", default=0,
- help="samplerate at which the file should be represented")
- parser.add_option("-B","--bufsize",
- action="store", dest="bufsize", default=512,
- metavar = "<size>", type='int',
- help="buffer size [default=512]")
- parser.add_option("-H","--hopsize",
- metavar = "<size>", type='int',
- action="store", dest="hopsize", default=256,
- help="overlap size [default=256]")
- parser.add_option("-t","--onset-threshold",
- metavar = "<value>", type="float",
- action="store", dest="threshold", default=0.3,
- help="onset peak picking threshold [default=0.3]")
- parser.add_option("-c","--cut",
- action="store_true", dest="cut", default=False,
- help="cut input sound file at detected labels \
- best used with option -L")
-
- # minioi
- parser.add_option("-M","--minioi",
- metavar = "<value>", type='string',
- action="store", dest="minioi", default="12ms",
- help="minimum inter onset interval [default=12ms]")
-
- """
- parser.add_option("-D","--delay",
- action = "store", dest = "delay", type = "float",
- metavar = "<seconds>", default=0,
- help="number of seconds to take back [default=system]\
- default system delay is 3*hopsize/samplerate")
- parser.add_option("-C","--dcthreshold",
- metavar = "<value>",
- action="store", dest="dcthreshold", default=1.,
- help="onset peak picking DC component [default=1.]")
- parser.add_option("-L","--localmin",
- action="store_true", dest="localmin", default=False,
- help="use local minima after peak detection")
- parser.add_option("-d","--derivate",
- action="store_true", dest="derivate", default=False,
- help="derivate onset detection function")
- parser.add_option("-z","--zerocross",
- metavar = "<value>",
- action="store", dest="zerothres", default=0.008,
- help="zero-crossing threshold for slicing [default=0.00008]")
- """
- # plotting functions
- """
- parser.add_option("-p","--plot",
- action="store_true", dest="plot", default=False,
- help="draw plot")
- parser.add_option("-x","--xsize",
- metavar = "<size>",
- action="store", dest="xsize", default=1.,
- type='float', help="define xsize for plot")
- parser.add_option("-y","--ysize",
- metavar = "<size>",
- action="store", dest="ysize", default=1.,
- type='float', help="define ysize for plot")
- parser.add_option("-f","--function",
- action="store_true", dest="func", default=False,
- help="print detection function")
- parser.add_option("-n","--no-onsets",
- action="store_true", dest="nplot", default=False,
- help="do not plot detected onsets")
- parser.add_option("-O","--outplot",
- metavar = "<output_image>",
- action="store", dest="outplot", default=None,
- help="save plot to output.{ps,png}")
- parser.add_option("-F","--spectrogram",
- action="store_true", dest="spectro", default=False,
- help="add spectrogram to the plot")
- """
- parser.add_option("-o","--output", type = str,
- metavar = "<outputdir>",
- action="store", dest="output_directory", default=None,
- help="specify path where slices of the original file should be created")
- parser.add_option("--cut-until-nsamples", type = int,
- metavar = "<samples>",
- action = "store", dest = "cut_until_nsamples", default = None,
- help="how many extra samples should be added at the end of each slice")
- parser.add_option("--cut-every-nslices", type = int,
- metavar = "<samples>",
- action = "store", dest = "cut_every_nslices", default = None,
- help="how many slices should be groupped together at each cut")
- parser.add_option("--cut-until-nslices", type = int,
- metavar = "<slices>",
- action = "store", dest = "cut_until_nslices", default = None,
- help="how many extra slices should be added at the end of each slice")
-
- parser.add_option("-v","--verbose",
- action="store_true", dest="verbose", default=True,
- help="make lots of noise [default]")
- parser.add_option("-q","--quiet",
- action="store_false", dest="verbose", default=True,
- help="be quiet")
- (options, args) = parser.parse_args()
- if not options.source_file:
- import os.path
- if len(args) == 1:
- options.source_file = args[0]
- else:
- print ("no file name given\n" + usage)
- sys.exit(1)
- return options, args
-
-if __name__ == '__main__':
- options, args = parse_args()
-
- hopsize = options.hopsize
- bufsize = options.bufsize
- samplerate = options.samplerate
- source_file = options.source_file
-
- from aubio import onset, tempo, source, sink
-
- s = source(source_file, samplerate, hopsize)
- if samplerate == 0: samplerate = s.get_samplerate()
-
- if options.beat:
- o = tempo(options.onset_method, bufsize, hopsize)
- else:
- o = onset(options.onset_method, bufsize, hopsize)
- if options.minioi:
- if options.minioi.endswith('ms'):
- o.set_minioi_ms(int(options.minioi[:-2]))
- elif options.minioi.endswith('s'):
- o.set_minioi_s(int(options.minioi[:-1]))
- else:
- o.set_minioi(int(options.minioi))
- o.set_threshold(options.threshold)
-
- timestamps = []
- total_frames = 0
- # analyze pass
- while True:
- samples, read = s()
- if o(samples):
- timestamps.append (o.get_last())
- if options.verbose: print ("%.4f" % o.get_last_s())
- total_frames += read
- if read < hopsize: break
- del s
- # print some info
- nstamps = len(timestamps)
- duration = float (total_frames) / float(samplerate)
- info = 'found %(nstamps)d timestamps in %(source_file)s' % locals()
- info += ' (total %(duration).2fs at %(samplerate)dHz)\n' % locals()
- sys.stderr.write(info)
-
- # cutting pass
- if options.cut and nstamps > 0:
- # generate output files
- from aubio.slicing import slice_source_at_stamps
- timestamps_end = None
- if options.cut_every_nslices:
- timestamps = timestamps[::options.cut_every_nslices]
- nstamps = len(timestamps)
- if options.cut_until_nslices and options.cut_until_nsamples:
- print ("warning: using cut_until_nslices, but cut_until_nsamples is set")
- if options.cut_until_nsamples:
- timestamps_end = [t + options.cut_until_nsamples for t in timestamps[1:]]
- timestamps_end += [ 1e120 ]
- if options.cut_until_nslices:
- timestamps_end = [t for t in timestamps[1 + options.cut_until_nslices:]]
- timestamps_end += [ 1e120 ] * (options.cut_until_nslices + 1)
- slice_source_at_stamps(source_file, timestamps, timestamps_end = timestamps_end,
- output_dir = options.output_directory,
- samplerate = samplerate)
-
- # print some info
- duration = float (total_frames) / float(samplerate)
- info = 'created %(nstamps)d slices from %(source_file)s' % locals()
- info += ' (total %(duration).2fs at %(samplerate)dHz)\n' % locals()
- sys.stderr.write(info)
--- a/setup.py
+++ b/setup.py
@@ -57,7 +57,6 @@
version = __version__,
packages = ['aubio'],
package_dir = {'aubio':'python/lib/aubio'},
- scripts = ['python/scripts/aubiocut', 'python/scripts/aubio'],
ext_modules = [aubio_extension],
description = 'a collection of tools for music analysis',
long_description = 'a collection of tools for music analysis',
@@ -75,6 +74,12 @@
'clean': CleanGenerated,
'build_ext': build_ext,
},
+ entry_points = {
+ 'console_scripts': [
+ 'aubio = aubio.cmd:main',
+ 'aubiocut = aubio.cut:main',
+ ],
+ },
test_suite = 'nose2.collector.collector',
extras_require = {
'tests': ['numpy'],