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(dest='command',
23 description="", metavar="<command>")
26 subparser = subparsers.add_parser('onset',
27 help='get onset times',
28 formatter_class = argparse.ArgumentDefaultsHelpFormatter)
29 parser_add_input(subparser)
30 parser_add_buf_hop_size(subparser)
31 helpstr = "onset novelty function"
32 helpstr += " <default|energy|hfc|complex|phase|specdiff|kl|mkl|specflux>"
33 parser_add_method(subparser, helpstr=helpstr)
34 parser_add_threshold(subparser)
35 parser_add_silence(subparser)
36 parser_add_minioi(subparser)
37 parser_add_time_format(subparser)
38 parser_add_verbose_help(subparser)
39 subparser.set_defaults(process=process_onset)
42 subparser = subparsers.add_parser('pitch',
43 help='extract fundamental frequency')
44 parser_add_input(subparser)
45 parser_add_buf_hop_size(subparser, buf_size=2048)
46 helpstr = "pitch detection method <default|yinfft|yin|mcomb|fcomb|schmitt>"
47 parser_add_method(subparser, helpstr=helpstr)
48 parser_add_threshold(subparser)
49 parser_add_silence(subparser)
50 parser_add_time_format(subparser)
51 parser_add_verbose_help(subparser)
52 subparser.set_defaults(process=process_pitch)
55 subparser = subparsers.add_parser('beat',
56 help='get locations of beats')
57 parser_add_input(subparser)
58 parser_add_buf_hop_size(subparser, buf_size=1024, hop_size=512)
59 parser_add_time_format(subparser)
60 parser_add_verbose_help(subparser)
61 subparser.set_defaults(process=process_tempo)
64 subparser = subparsers.add_parser('notes',
65 help='get midi-like notes')
66 parser_add_input(subparser)
67 parser_add_buf_hop_size(subparser)
68 parser_add_time_format(subparser)
69 parser_add_verbose_help(subparser)
70 subparser.set_defaults(process=process_notes)
73 subparser = subparsers.add_parser('mfcc',
74 help='extract mel-frequency cepstrum coefficients')
75 parser_add_input(subparser)
76 parser_add_buf_hop_size(subparser)
77 parser_add_time_format(subparser)
78 parser_add_verbose_help(subparser)
79 subparser.set_defaults(process=process_mfcc)
82 subparser = subparsers.add_parser('melbands',
83 help='extract mel-frequency energies per band')
84 parser_add_input(subparser)
85 parser_add_buf_hop_size(subparser)
86 parser_add_time_format(subparser)
87 parser_add_verbose_help(subparser)
88 subparser.set_defaults(process=process_melbands)
92 def parser_add_input(parser):
93 parser.add_argument("source_uri", default=None, nargs='?',
94 help="input sound file to analyse", metavar = "<source_uri>")
95 parser.add_argument("-i", "--input", dest = "source_uri2",
96 help="input sound file to analyse", metavar = "<source_uri>")
97 parser.add_argument("-r", "--samplerate",
98 metavar = "<freq>", type=int,
99 action="store", dest="samplerate", default=0,
100 help="samplerate at which the file should be represented")
102 def parser_add_verbose_help(parser):
103 parser.add_argument("-v","--verbose",
104 action="count", dest="verbose", default=1,
105 help="make lots of noise [default]")
106 parser.add_argument("-q","--quiet",
107 action="store_const", dest="verbose", const=0,
110 def parser_add_buf_hop_size(parser, buf_size=512, hop_size=256):
111 parser.add_argument("-B","--bufsize",
112 action="store", dest="buf_size", default=buf_size,
113 metavar = "<size>", type=int,
114 help="buffer size [default=%d]" % buf_size)
115 parser.add_argument("-H","--hopsize",
116 metavar = "<size>", type=int,
117 action="store", dest="hop_size", default=hop_size,
118 help="overlap size [default=%d]" % hop_size)
120 def parser_add_method(parser, method='default', helpstr='method'):
121 parser.add_argument("-m","--method",
122 metavar = "<method>", type=str,
123 action="store", dest="method", default=method,
124 help="%s [default=%s]" % (helpstr, method))
126 def parser_add_threshold(parser, default=None):
127 parser.add_argument("-t","--threshold",
128 metavar = "<threshold>", type=float,
129 action="store", dest="threshold", default=default,
130 help="threshold [default=%s]" % default)
132 def parser_add_silence(parser):
133 parser.add_argument("-s", "--silence",
134 metavar = "<value>", type=float,
135 action="store", dest="silence", default=-70,
136 help="silence threshold")
138 def parser_add_minioi(parser):
139 parser.add_argument("-M", "--minioi",
140 metavar = "<value>", type=str,
141 action="store", dest="minioi", default="12ms",
142 help="minimum Inter-Onset Interval")
144 def parser_add_time_format(parser):
145 helpstr = "select time values output format (samples, ms, seconds)"
146 helpstr += " [default=seconds]"
147 parser.add_argument("-T", "--time-format",
155 def parse_options(args, valid_opts):
156 options = {k :v for k,v in vars(args).items() if k in valid_opts}
159 def remap_pvoc_options(options):
160 # remap buf_size to win_s, hop_size to hop_s
161 # FIXME: adjust python/ext/py-phasevoc.c to understand buf_size/hop_size
162 options['win_s'] = options['buf_size']
163 del options['buf_size']
164 options['hop_s'] = options['hop_size']
165 del options['hop_size']
168 def samples2seconds(n_frames, samplerate):
169 return "%f\t" % (n_frames / float(samplerate))
171 def samples2milliseconds(n_frames, samplerate):
172 return "%f\t" % (1000. * n_frames / float(samplerate))
174 def samples2samples(n_frames, samplerate):
175 return "%d\t" % n_frames
178 if mode is None or mode == 'seconds' or mode == 's':
179 return samples2seconds
180 elif mode == 'ms' or mode == 'milliseconds':
181 return samples2milliseconds
182 elif mode == 'samples':
183 return samples2samples
185 raise ValueError('invalid time format %s' % mode)
187 # definition of processing classes
189 class default_process(object):
190 def __init__(self, args):
191 if 'time_format' in args:
192 self.time2string = timefunc(args.time_format)
193 if args.verbose > 2 and hasattr(self, 'options'):
194 name = type(self).__name__.split('_')[1]
195 optstr = ' '.join(['running', name, 'with options', repr(self.options), '\n'])
196 sys.stderr.write(optstr)
197 def flush(self, n_frames, samplerate):
200 class process_onset(default_process):
201 valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
202 def __init__(self, args):
203 self.options = parse_options(args, self.valid_opts)
204 self.onset = aubio.onset(**self.options)
205 if args.threshold is not None:
206 self.onset.set_threshold(args.threshold)
208 if args.minioi.endswith('ms'):
209 self.onset.set_minioi_ms(float(args.minioi[:-2]))
210 elif args.minioi.endswith('s'):
211 self.onset.set_minioi_s(float(args.minioi[:-1]))
213 self.onset.set_minioi(int(args.minioi))
215 self.onset.set_silence(args.silence)
216 super(process_onset, self).__init__(args)
217 def __call__(self, block):
218 return self.onset(block)
219 def repr_res(self, res, frames_read, samplerate):
221 outstr = self.time2string(self.onset.get_last(), samplerate)
222 sys.stdout.write(outstr + '\n')
224 class process_pitch(default_process):
225 valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
226 def __init__(self, args):
227 self.options = parse_options(args, self.valid_opts)
228 self.pitch = aubio.pitch(**self.options)
229 if args.threshold is not None:
230 self.pitch.set_tolerance(args.threshold)
231 if args.silence is not None:
232 self.pitch.set_silence(args.silence)
233 super(process_pitch, self).__init__(args)
234 def __call__(self, block):
235 return self.pitch(block)
236 def repr_res(self, res, frames_read, samplerate):
237 fmt_out = self.time2string(frames_read, samplerate)
238 sys.stdout.write(fmt_out + "%.6f\n" % res[0])
240 class process_tempo(default_process):
241 valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
242 def __init__(self, args):
243 self.options = parse_options(args, self.valid_opts)
244 self.tempo = aubio.tempo(**self.options)
245 super(process_tempo, self).__init__(args)
246 def __call__(self, block):
247 return self.tempo(block)
248 def repr_res(self, res, frames_read, samplerate):
250 outstr = self.time2string(self.tempo.get_last(), samplerate)
251 sys.stdout.write(outstr + '\n')
253 class process_notes(default_process):
254 valid_opts = ['method', 'hop_size', 'buf_size', 'samplerate']
255 def __init__(self, args):
256 self.options = parse_options(args, self.valid_opts)
257 self.notes = aubio.notes(**self.options)
258 super(process_notes, self).__init__(args)
259 def __call__(self, block):
260 return self.notes(block)
261 def repr_res(self, res, frames_read, samplerate):
262 if res[2] != 0: # note off
263 fmt_out = self.time2string(frames_read, samplerate)
264 sys.stdout.write(fmt_out + '\n')
265 if res[0] != 0: # note on
267 fmt_out = "%f\t" % lastmidi
268 fmt_out += self.time2string(frames_read, samplerate)
269 sys.stdout.write(fmt_out) # + '\t')
270 def flush(self, frames_read, samplerate):
271 eof = self.time2string(frames_read, samplerate)
272 sys.stdout.write(eof + '\n')
274 class process_mfcc(default_process):
275 def __init__(self, args):
276 valid_opts = ['hop_size', 'buf_size']
277 options = parse_options(args, valid_opts)
278 self.options = remap_pvoc_options(options)
279 self.pv = aubio.pvoc(**options)
281 valid_opts = ['buf_size', 'n_filters', 'n_coeffs', 'samplerate']
282 options = parse_options(args, valid_opts)
283 self.mfcc = aubio.mfcc(**options)
284 self.options.update(options)
286 super(process_mfcc, self).__init__(args)
288 def __call__(self, block):
289 fftgrain = self.pv(block)
290 return self.mfcc(fftgrain)
291 def repr_res(self, res, frames_read, samplerate):
292 fmt_out = self.time2string(frames_read, samplerate)
293 fmt_out += ' '.join(["% 9.7f" % f for f in res.tolist()])
294 sys.stdout.write(fmt_out + '\n')
296 class process_melbands(default_process):
297 def __init__(self, args):
299 valid_opts = ['hop_size', 'buf_size']
300 options = parse_options(args, valid_opts)
301 options = remap_pvoc_options(options)
302 self.pv = aubio.pvoc(**options)
304 valid_opts = ['buf_size', 'n_filters']
305 options = {k :v for k,v in vars(args).items() if k in valid_opts}
307 options['win_s'] = options['buf_size']
308 del options['buf_size']
309 self.filterbank = aubio.filterbank(**options)
310 self.filterbank.set_mel_coeffs_slaney(args.samplerate)
312 super(process_melbands, self).__init__(args)
313 def __call__(self, block):
314 fftgrain = self.pv(block)
315 return self.filterbank(fftgrain)
316 def repr_res(self, res, frames_read, samplerate):
317 fmt_out = self.time2string(frames_read, samplerate)
318 fmt_out += ' '.join(["% 9.7f" % f for f in res.tolist()])
319 sys.stdout.write(fmt_out + '\n')
322 parser = aubio_parser()
323 args = parser.parse_args()
324 if args.show_version or ('verbose' in args and args.verbose > 3):
325 sys.stdout.write('aubio version ' + aubio.version + '\n')
326 if args.show_version and args.command is None:
328 if args.command is None:
329 sys.stderr.write("Error: a command is required\n")
332 elif not args.source_uri and not args.source_uri2:
333 sys.stderr.write("Error: a source is required\n")
336 elif args.source_uri2 is not None:
337 args.source_uri = args.source_uri2
340 with aubio.source(args.source_uri, hop_size=args.hop_size,
341 samplerate=args.samplerate) as a_source:
342 args.samplerate = a_source.samplerate
343 # create the processor for this subcommand
344 processor = args.process(args)
347 # read new block from source
348 block, read = a_source()
349 # execute processor on this block
350 res = processor(block)
351 # print results for this block
353 processor.repr_res(res, frames_read, a_source.samplerate)
354 # increment total number of frames read
356 # exit loop at end of file
357 if read < a_source.hop_size: break
358 # flush the processor if needed
359 processor.flush(frames_read, a_source.samplerate)
361 fmt_string = "read {:.2f}s"
362 fmt_string += " ({:d} samples in {:d} blocks of {:d})"
363 fmt_string += " from {:s} at {:d}Hz\n"
364 sys.stderr.write(fmt_string.format(
365 frames_read/float(a_source.samplerate),
367 frames_read // a_source.hop_size + 1,
370 a_source.samplerate))
371 except KeyboardInterrupt as e: