source: python/lib/aubio/slicing.py @ c9c012e

feature/autosinkfeature/cnnfeature/cnn_orgfeature/constantqfeature/crepefeature/crepe_orgfeature/pitchshiftfeature/pydocstringsfeature/timestretchfix/ffmpeg5
Last change on this file since c9c012e was c9c012e, checked in by Paul Brossier <piem@piem.org>, 5 years ago

[py] [slicing] add option create_first, default to False

  • Property mode set to 100644
File size: 3.8 KB
Line 
1"""utility routines to slice sound files at given timestamps"""
2
3import os
4from aubio import source, sink
5
6_max_timestamp = 1e120
7
8def slice_source_at_stamps(source_file, timestamps, timestamps_end=None,
9                           output_dir=None, samplerate=0, hopsize=256,
10                           create_first=False):
11    """ slice a sound file at given timestamps """
12
13    if timestamps is None or len(timestamps) == 0:
14        raise ValueError("no timestamps given")
15
16    if timestamps[0] != 0 and create_first:
17        timestamps = [0] + timestamps
18        if timestamps_end is not None:
19            timestamps_end = [timestamps[1] - 1] + timestamps_end
20
21    if timestamps_end is not None:
22        if len(timestamps_end) == len(timestamps) - 1:
23            timestamps_end = timestamps_end + [_max_timestamp]
24        elif len(timestamps_end) != len(timestamps):
25            raise ValueError("len(timestamps_end) != len(timestamps)")
26    else:
27        timestamps_end = [t - 1 for t in timestamps[1:]] + [_max_timestamp]
28
29    regions = list(zip(timestamps, timestamps_end))
30    #print regions
31
32    source_base_name, _ = os.path.splitext(os.path.basename(source_file))
33    if output_dir is not None:
34        if not os.path.isdir(output_dir):
35            os.makedirs(output_dir)
36        source_base_name = os.path.join(output_dir, source_base_name)
37
38    def new_sink_name(source_base_name, timestamp, samplerate):
39        """ create a sink based on a timestamp in samples, converted in seconds """
40        timestamp_seconds = timestamp / float(samplerate)
41        return source_base_name + "_%011.6f" % timestamp_seconds + '.wav'
42
43    # open source file
44    _source = source(source_file, samplerate, hopsize)
45    samplerate = _source.samplerate
46
47    total_frames = 0
48    slices = []
49
50    while True:
51        # get hopsize new samples from source
52        vec, read = _source.do_multi()
53        # if the total number of frames read will exceed the next region start
54        while len(regions) and total_frames + read >= regions[0][0]:
55            #print "getting", regions[0], "at", total_frames
56            # get next region
57            start_stamp, end_stamp = regions.pop(0)
58            # create a name for the sink
59            new_sink_path = new_sink_name(source_base_name, start_stamp, samplerate)
60            # create its sink
61            _sink = sink(new_sink_path, samplerate, _source.channels)
62            # create a dictionary containing all this
63            new_slice = {'start_stamp': start_stamp, 'end_stamp': end_stamp, 'sink': _sink}
64            # append the dictionary to the current list of slices
65            slices.append(new_slice)
66
67        for current_slice in slices:
68            start_stamp = current_slice['start_stamp']
69            end_stamp = current_slice['end_stamp']
70            _sink = current_slice['sink']
71            # sample index to start writing from new source vector
72            start = max(start_stamp - total_frames, 0)
73            # number of samples yet to written be until end of region
74            remaining = end_stamp - total_frames + 1
75            #print current_slice, remaining, start
76            # not enough frames remaining, time to split
77            if remaining < read:
78                if remaining > start:
79                    # write remaining samples from current region
80                    _sink.do_multi(vec[:, start:remaining], remaining - start)
81                    #print("closing region", "remaining", remaining)
82                    # close this file
83                    _sink.close()
84            elif read > start:
85                # write all the samples
86                _sink.do_multi(vec[:, start:read], read - start)
87        total_frames += read
88        # remove old slices
89        slices = list(filter(lambda s: s['end_stamp'] > total_frames,
90            slices))
91        if read < hopsize:
92            break
Note: See TracBrowser for help on using the repository browser.