2 # -*- coding: utf-8 -*-
4 """aubio command line tool
6 This file was written by Paul Brossier <piem@aubio.org> and is released under
9 Note: this script is mostly about parsing command line arguments. For more
10 readable code examples, check out the `python/demos` folder."""
17 epilog = 'use "%(prog)s <command> --help" for more info about each command'
18 parser = argparse.ArgumentParser(epilog=epilog)
19 parser.add_argument('-V', '--version', help="show version",
20 action="store_true", dest="show_version")
22 subparsers = parser.add_subparsers(title='commands', dest='command',
23 parser_class= AubioArgumentParser,
26 parser_add_subcommand_help(subparsers)
28 parser_add_subcommand_onset(subparsers)
29 parser_add_subcommand_pitch(subparsers)
30 parser_add_subcommand_beat(subparsers)
31 parser_add_subcommand_tempo(subparsers)
32 parser_add_subcommand_notes(subparsers)
33 parser_add_subcommand_mfcc(subparsers)
34 parser_add_subcommand_melbands(subparsers)
35 parser_add_subcommand_quiet(subparsers)
36 parser_add_subcommand_cut(subparsers)
40 def parser_add_subcommand_help(subparsers):
41 # global help subcommand
42 subparsers.add_parser('help',
43 help='show help message',
44 formatter_class = argparse.ArgumentDefaultsHelpFormatter)
46 def parser_add_subcommand_onset(subparsers):
48 subparser = subparsers.add_parser('onset',
49 help='estimate time of onsets (beginning of sound event)',
50 formatter_class = argparse.ArgumentDefaultsHelpFormatter)
52 subparser.add_buf_hop_size()
53 helpstr = "onset novelty function"
54 helpstr += " <default|energy|hfc|complex|phase|specdiff|kl|mkl|specflux>"
55 subparser.add_method(helpstr=helpstr)
56 subparser.add_threshold()
57 subparser.add_silence()
58 subparser.add_minioi()
59 subparser.add_time_format()
60 subparser.add_verbose_help()
61 subparser.set_defaults(process=process_onset)
63 def parser_add_subcommand_pitch(subparsers):
65 subparser = subparsers.add_parser('pitch',
66 help='estimate fundamental frequency (monophonic)')
68 subparser.add_buf_hop_size(buf_size=2048)
69 helpstr = "pitch detection method <default|yinfft|yin|mcomb|fcomb|schmitt>"
70 subparser.add_method(helpstr=helpstr)
71 subparser.add_threshold()
72 subparser.add_pitch_unit()
73 subparser.add_silence()
74 subparser.add_time_format()
75 subparser.add_verbose_help()
76 subparser.set_defaults(process=process_pitch)
78 def parser_add_subcommand_beat(subparsers):
80 subparser = subparsers.add_parser('beat',
81 help='estimate location of beats')
83 subparser.add_buf_hop_size(buf_size=1024, hop_size=512)
84 subparser.add_time_format()
85 subparser.add_verbose_help()
86 subparser.set_defaults(process=process_beat)
88 def parser_add_subcommand_tempo(subparsers):
90 subparser = subparsers.add_parser('tempo',
91 help='estimate overall tempo in bpm')
93 subparser.add_buf_hop_size(buf_size=1024, hop_size=512)
94 subparser.add_time_format()
95 subparser.add_verbose_help()
96 subparser.set_defaults(process=process_tempo)
98 def parser_add_subcommand_notes(subparsers):
100 subparser = subparsers.add_parser('notes',
101 help='estimate midi-like notes (monophonic)')
102 subparser.add_input()
103 subparser.add_buf_hop_size()
104 subparser.add_time_format()
105 subparser.add_verbose_help()
106 subparser.set_defaults(process=process_notes)
108 def parser_add_subcommand_mfcc(subparsers):
110 subparser = subparsers.add_parser('mfcc',
111 help='extract Mel-Frequency Cepstrum Coefficients')
112 subparser.add_input()
113 subparser.add_buf_hop_size()
114 subparser.add_time_format()
115 subparser.add_verbose_help()
116 subparser.set_defaults(process=process_mfcc)
118 def parser_add_subcommand_melbands(subparsers):
119 # melbands subcommand
120 subparser = subparsers.add_parser('melbands',
121 help='extract energies in Mel-frequency bands')
122 subparser.add_input()
123 subparser.add_buf_hop_size()
124 subparser.add_time_format()
125 subparser.add_verbose_help()
126 subparser.set_defaults(process=process_melbands)
128 def parser_add_subcommand_quiet(subparsers):
130 subparser = subparsers.add_parser('quiet',
131 help='extract timestamps of quiet and loud regions')
132 subparser.add_input()
133 subparser.add_hop_size()
134 subparser.add_silence()
135 subparser.add_time_format()
136 subparser.add_verbose_help()
137 subparser.set_defaults(process=process_quiet)
139 def parser_add_subcommand_cut(subparsers):
141 subparser = subparsers.add_parser('cut',
142 help='slice at timestamps')
143 subparser.add_input()
144 helpstr = "onset novelty function"
145 helpstr += " <default|energy|hfc|complex|phase|specdiff|kl|mkl|specflux>"
146 subparser.add_method(helpstr=helpstr)
147 subparser.add_buf_hop_size()
148 subparser.add_silence()
149 subparser.add_threshold(default=0.3)
150 subparser.add_minioi()
151 subparser.add_slicer_options()
152 subparser.add_time_format()
153 subparser.add_verbose_help()
154 subparser.set_defaults(process=process_cut)
156 class AubioArgumentParser(argparse.ArgumentParser):
159 self.add_argument("source_uri", default=None, nargs='?',
160 help="input sound file to analyse", metavar = "<source_uri>")
161 self.add_argument("-i", "--input", dest = "source_uri2",
162 help="input sound file to analyse", metavar = "<source_uri>")
163 self.add_argument("-r", "--samplerate",
164 metavar = "<freq>", type=int,
165 action="store", dest="samplerate", default=0,
166 help="samplerate at which the file should be represented")
168 def add_verbose_help(self):
169 self.add_argument("-v","--verbose",
170 action="count", dest="verbose", default=1,
171 help="make lots of noise [default]")
172 self.add_argument("-q","--quiet",
173 action="store_const", dest="verbose", const=0,
176 def add_buf_hop_size(self, buf_size=512, hop_size=256):
177 self.add_buf_size(buf_size=buf_size)
178 self.add_hop_size(hop_size=hop_size)
180 def add_buf_size(self, buf_size=512):
181 self.add_argument("-B","--bufsize",
182 action="store", dest="buf_size", default=buf_size,
183 metavar = "<size>", type=int,
184 help="buffer size [default=%d]" % buf_size)
186 def add_hop_size(self, hop_size=256):
187 self.add_argument("-H","--hopsize",
188 metavar = "<size>", type=int,
189 action="store", dest="hop_size", default=hop_size,
190 help="overlap size [default=%d]" % hop_size)
192 def add_method(self, method='default', helpstr='method'):
193 self.add_argument("-m","--method",
194 metavar = "<method>", type=str,
195 action="store", dest="method", default=method,
196 help="%s [default=%s]" % (helpstr, method))
198 def add_threshold(self, default=None):
199 self.add_argument("-t","--threshold",
200 metavar = "<threshold>", type=float,
201 action="store", dest="threshold", default=default,
202 help="threshold [default=%s]" % default)
204 def add_silence(self):
205 self.add_argument("-s", "--silence",
206 metavar = "<value>", type=float,
207 action="store", dest="silence", default=-70,
208 help="silence threshold")
210 def add_minioi(self, default="12ms"):
211 self.add_argument("-M", "--minioi",
212 metavar = "<value>", type=str,
213 action="store", dest="minioi", default=default,
214 help="minimum Inter-Onset Interval [default=%s]" % default)
216 def add_pitch_unit(self, default="Hz"):
217 help_str = "frequency unit, should be one of Hz, midi, bin, cent"
218 help_str += " [default=%s]" % default
219 self.add_argument("-u", "--pitch-unit",
220 metavar = "<value>", type=str,
221 action="store", dest="pitch_unit", default=default,
224 def add_time_format(self):
225 helpstr = "select time values output format (samples, ms, seconds)"
226 helpstr += " [default=seconds]"
227 self.add_argument("-T", "--time-format",
233 def add_slicer_options(self):
234 self.add_argument("-o","--output", type = str,
235 metavar = "<outputdir>",
236 action="store", dest="output_directory", default=None,
237 help="specify path where slices of the original file should be created")
238 self.add_argument("--cut-until-nsamples", type = int,
239 metavar = "<samples>",
240 action = "store", dest = "cut_until_nsamples", default = None,
241 help="how many extra samples should be added at the end of each slice")
242 self.add_argument("--cut-every-nslices", type = int,
243 metavar = "<samples>",
244 action = "store", dest = "cut_every_nslices", default = None,
245 help="how many slices should be groupped together at each cut")
246 self.add_argument("--cut-until-nslices", type = int,
247 metavar = "<slices>",
248 action = "store", dest = "cut_until_nslices", default = None,
249 help="how many extra slices should be added at the end of each slice")
253 def samples2seconds(n_frames, samplerate):
254 return "%f\t" % (n_frames / float(samplerate))
256 def samples2milliseconds(n_frames, samplerate):
257 return "%f\t" % (1000. * n_frames / float(samplerate))
259 def samples2samples(n_frames, _samplerate):
260 return "%d\t" % n_frames
263 if mode is None or mode == 'seconds' or mode == 's':
264 return samples2seconds
265 elif mode == 'ms' or mode == 'milliseconds':
266 return samples2milliseconds
267 elif mode == 'samples':
268 return samples2samples
270 raise ValueError("invalid time format '%s'" % mode)
272 # definition of processing classes
274 class default_process(object):
275 def __init__(self, args):
276 if 'time_format' in args:
277 self.time2string = timefunc(args.time_format)
278 if args.verbose > 2 and hasattr(self, 'options'):
279 name = type(self).__name__.split('_')[1]
280 optstr = ' '.join(['running', name, 'with options', repr(self.options), '\n'])
281 sys.stderr.write(optstr)
282 def flush(self, frames_read, samplerate):
283 # optionally called at the end of process
286 def parse_options(self, args, valid_opts):
287 # get any valid options found in a dictionnary of arguments
288 options = {k :v for k,v in vars(args).items() if k in valid_opts}
289 self.options = options
291 def remap_pvoc_options(self, options):
292 # FIXME: we need to remap buf_size to win_s, hop_size to hop_s
293 # adjust python/ext/py-phasevoc.c to understand buf_size/hop_size
294 if 'buf_size' in options:
295 options['win_s'] = options['buf_size']
296 del options['buf_size']
297 if 'hop_size' in options:
298 options['hop_s'] = options['hop_size']
299 del options['hop_size']
300 self.options = options
302 class process_onset(default_process):
303 valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
304 def __init__(self, args):
305 self.parse_options(args, self.valid_opts)
306 self.onset = aubio.onset(**self.options)
307 if args.threshold is not None:
308 self.onset.set_threshold(args.threshold)
310 if args.minioi.endswith('ms'):
311 self.onset.set_minioi_ms(float(args.minioi[:-2]))
312 elif args.minioi.endswith('s'):
313 self.onset.set_minioi_s(float(args.minioi[:-1]))
315 self.onset.set_minioi(int(args.minioi))
317 self.onset.set_silence(args.silence)
318 super(process_onset, self).__init__(args)
319 def __call__(self, block):
320 return self.onset(block)
321 def repr_res(self, res, _frames_read, samplerate):
323 outstr = self.time2string(self.onset.get_last(), samplerate)
324 sys.stdout.write(outstr + '\n')
326 class process_pitch(default_process):
327 valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
328 def __init__(self, args):
329 self.parse_options(args, self.valid_opts)
330 self.pitch = aubio.pitch(**self.options)
331 if args.pitch_unit is not None:
332 self.pitch.set_unit(args.pitch_unit)
333 if args.threshold is not None:
334 self.pitch.set_tolerance(args.threshold)
335 if args.silence is not None:
336 self.pitch.set_silence(args.silence)
337 super(process_pitch, self).__init__(args)
338 def __call__(self, block):
339 return self.pitch(block)
340 def repr_res(self, res, frames_read, samplerate):
341 fmt_out = self.time2string(frames_read, samplerate)
342 sys.stdout.write(fmt_out + "%.6f\n" % res[0])
344 class process_beat(default_process):
345 valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
346 def __init__(self, args):
347 self.parse_options(args, self.valid_opts)
348 self.tempo = aubio.tempo(**self.options)
349 super(process_beat, self).__init__(args)
350 def __call__(self, block):
351 return self.tempo(block)
352 def repr_res(self, res, _frames_read, samplerate):
354 outstr = self.time2string(self.tempo.get_last(), samplerate)
355 sys.stdout.write(outstr + '\n')
357 class process_tempo(process_beat):
358 def __init__(self, args):
359 super(process_tempo, self).__init__(args)
360 self.beat_locations = []
361 def repr_res(self, res, _frames_read, samplerate):
363 self.beat_locations.append(self.tempo.get_last_s())
364 def flush(self, frames_read, samplerate):
366 if len(self.beat_locations) < 2:
367 outstr = "unknown bpm"
369 bpms = 60./ np.diff(self.beat_locations)
370 median_bpm = np.mean(bpms)
371 if len(self.beat_locations) < 10:
372 outstr = "%.2f bpm (uncertain)" % median_bpm
374 outstr = "%.2f bpm" % median_bpm
375 sys.stdout.write(outstr + '\n')
377 class process_notes(default_process):
378 valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
379 def __init__(self, args):
380 self.parse_options(args, self.valid_opts)
381 self.notes = aubio.notes(**self.options)
382 super(process_notes, self).__init__(args)
383 def __call__(self, block):
384 return self.notes(block)
385 def repr_res(self, res, frames_read, samplerate):
386 if res[2] != 0: # note off
387 fmt_out = self.time2string(frames_read, samplerate)
388 sys.stdout.write(fmt_out + '\n')
389 if res[0] != 0: # note on
391 fmt_out = "%f\t" % lastmidi
392 fmt_out += self.time2string(frames_read, samplerate)
393 sys.stdout.write(fmt_out) # + '\t')
394 def flush(self, frames_read, samplerate):
395 eof = self.time2string(frames_read, samplerate)
396 sys.stdout.write(eof + '\n')
398 class process_mfcc(default_process):
399 def __init__(self, args):
400 valid_opts1 = ['hop_size', 'buf_size']
401 self.parse_options(args, valid_opts1)
402 self.remap_pvoc_options(self.options)
403 self.pv = aubio.pvoc(**self.options)
405 valid_opts2 = ['buf_size', 'n_filters', 'n_coeffs', 'samplerate']
406 self.parse_options(args, valid_opts2)
407 self.mfcc = aubio.mfcc(**self.options)
409 # remember all options
410 self.parse_options(args, list(set(valid_opts1 + valid_opts2)))
412 super(process_mfcc, self).__init__(args)
414 def __call__(self, block):
415 fftgrain = self.pv(block)
416 return self.mfcc(fftgrain)
417 def repr_res(self, res, frames_read, samplerate):
418 fmt_out = self.time2string(frames_read, samplerate)
419 fmt_out += ' '.join(["% 9.7f" % f for f in res.tolist()])
420 sys.stdout.write(fmt_out + '\n')
422 class process_melbands(default_process):
423 def __init__(self, args):
425 valid_opts = ['hop_size', 'buf_size']
426 self.parse_options(args, valid_opts)
427 self.remap_pvoc_options(self.options)
428 self.pv = aubio.pvoc(**self.options)
430 valid_opts = ['buf_size', 'n_filters']
431 self.parse_options(args, valid_opts)
432 self.remap_pvoc_options(self.options)
433 self.filterbank = aubio.filterbank(**self.options)
434 self.filterbank.set_mel_coeffs_slaney(args.samplerate)
436 super(process_melbands, self).__init__(args)
437 def __call__(self, block):
438 fftgrain = self.pv(block)
439 return self.filterbank(fftgrain)
440 def repr_res(self, res, frames_read, samplerate):
441 fmt_out = self.time2string(frames_read, samplerate)
442 fmt_out += ' '.join(["% 9.7f" % f for f in res.tolist()])
443 sys.stdout.write(fmt_out + '\n')
445 class process_quiet(default_process):
446 def __init__(self, args):
448 valid_opts = ['hop_size', 'silence']
449 self.parse_options(args, valid_opts)
452 if args.silence is not None:
453 self.silence = args.silence
454 super(process_quiet, self).__init__(args)
456 def __call__(self, block):
457 if aubio.silence_detection(block, self.silence) == 1:
458 if self.wassilence != 1:
460 return 2 # newly found silence
461 return 1 # silence again
463 if self.wassilence != 0:
465 return -1 # newly found noise
466 return 0 # noise again
468 def repr_res(self, res, frames_read, samplerate):
474 if fmt_out is not None:
475 fmt_out += self.time2string(frames_read, samplerate)
476 sys.stdout.write(fmt_out + '\n')
478 class process_cut(process_onset):
479 def __init__(self, args):
480 super(process_cut, self).__init__(args)
484 def __call__(self, block):
485 ret = super(process_cut, self).__call__(block)
486 if ret: self.slices.append(self.onset.get_last())
489 def flush(self, frames_read, samplerate):
490 from aubio.cut import _cut_slice
491 _cut_slice(self.options, self.slices)
492 duration = float (frames_read) / float(samplerate)
493 base_info = '%(source_file)s' % {'source_file': self.options.source_uri}
494 base_info += ' (total %(duration).2fs at %(samplerate)dHz)\n' % \
495 {'duration': duration, 'samplerate': samplerate}
496 info = "created %d slices from " % len(self.slices)
498 sys.stderr.write(info)
501 parser = aubio_parser()
502 args = parser.parse_args()
503 if 'show_version' in args and args.show_version:
504 sys.stdout.write('aubio version ' + aubio.version + '\n')
506 elif 'verbose' in args and args.verbose > 3:
507 sys.stderr.write('aubio version ' + aubio.version + '\n')
508 if 'command' not in args or args.command is None or args.command in ['help']:
509 # no command given, print help and return 1
511 if args.command and args.command in ['help']:
515 elif not args.source_uri and not args.source_uri2:
516 sys.stderr.write("Error: a source is required\n")
519 elif args.source_uri2 is not None:
520 args.source_uri = args.source_uri2
523 with aubio.source(args.source_uri, hop_size=args.hop_size,
524 samplerate=args.samplerate) as a_source:
525 # always update args.samplerate to native samplerate, in case
526 # source was opened with args.samplerate=0
527 args.samplerate = a_source.samplerate
528 # create the processor for this subcommand
529 processor = args.process(args)
532 # read new block from source
533 block, read = a_source()
534 # execute processor on this block
535 res = processor(block)
536 # print results for this block
538 processor.repr_res(res, frames_read, a_source.samplerate)
539 # increment total number of frames read
541 # exit loop at end of file
542 if read < a_source.hop_size: break
543 # flush the processor if needed
544 processor.flush(frames_read, a_source.samplerate)
546 fmt_string = "read {:.2f}s"
547 fmt_string += " ({:d} samples in {:d} blocks of {:d})"
548 fmt_string += " from {:s} at {:d}Hz\n"
549 sys.stderr.write(fmt_string.format(
550 frames_read/float(a_source.samplerate),
552 frames_read // a_source.hop_size + 1,
555 a_source.samplerate))
556 except KeyboardInterrupt: