source: python/lib/aubio/slicing.py @ 60c8a73

feature/autosinkfeature/cnnfeature/cnn_orgfeature/constantqfeature/crepefeature/crepe_orgfeature/pitchshiftfeature/pydocstringsfeature/timestretchfix/ffmpeg5
Last change on this file since 60c8a73 was 60c8a73, checked in by Paul Brossier <piem@piem.org>, 5 years ago

[py] slicing.py: fix for slices smaller than hop_size

  • Property mode set to 100644
File size: 3.6 KB
RevLine 
[8b56b18]1"""utility routines to slice sound files at given timestamps"""
2
[88432a9]3import os
[8b56b18]4from aubio import source, sink
[88432a9]5
[8b56b18]6_max_timestamp = 1e120
[f36ecea]7
[8b56b18]8def slice_source_at_stamps(source_file, timestamps, timestamps_end=None,
9                           output_dir=None, samplerate=0, hopsize=256):
10    """ slice a sound file at given timestamps """
[88432a9]11
[8b56b18]12    if timestamps is None or len(timestamps) == 0:
13        raise ValueError("no timestamps given")
[aee840b]14
[f36ecea]15    if timestamps[0] != 0:
16        timestamps = [0] + timestamps
[dc654f8]17        if timestamps_end is not None:
[35a44e9]18            timestamps_end = [timestamps[1] - 1] + timestamps_end
[f36ecea]19
[dc654f8]20    if timestamps_end is not None:
[4320679]21        if len(timestamps_end) != len(timestamps):
[8b56b18]22            raise ValueError("len(timestamps_end) != len(timestamps)")
[f36ecea]23    else:
[8b56b18]24        timestamps_end = [t - 1 for t in timestamps[1:]] + [_max_timestamp]
[aee840b]25
[0e59ae0]26    regions = list(zip(timestamps, timestamps_end))
[4320679]27    #print regions
28
[8b56b18]29    source_base_name, _ = os.path.splitext(os.path.basename(source_file))
[dc654f8]30    if output_dir is not None:
[88432a9]31        if not os.path.isdir(output_dir):
32            os.makedirs(output_dir)
33        source_base_name = os.path.join(output_dir, source_base_name)
34
[f36ecea]35    def new_sink_name(source_base_name, timestamp, samplerate):
[8b56b18]36        """ create a sink based on a timestamp in samples, converted in seconds """
[f36ecea]37        timestamp_seconds = timestamp / float(samplerate)
[6fbee46]38        return source_base_name + "_%011.6f" % timestamp_seconds + '.wav'
[88432a9]39
[8b56b18]40    # open source file
41    _source = source(source_file, samplerate, hopsize)
[1b62ee9]42    samplerate = _source.samplerate
[f36ecea]43
[4320679]44    total_frames = 0
45    slices = []
[88432a9]46
47    while True:
48        # get hopsize new samples from source
[8b56b18]49        vec, read = _source.do_multi()
[4320679]50        # if the total number of frames read will exceed the next region start
[60c8a73]51        while len(regions) and total_frames + read >= regions[0][0]:
[4320679]52            #print "getting", regions[0], "at", total_frames
53            # get next region
54            start_stamp, end_stamp = regions.pop(0)
55            # create a name for the sink
[f36ecea]56            new_sink_path = new_sink_name(source_base_name, start_stamp, samplerate)
[4320679]57            # create its sink
[8b56b18]58            _sink = sink(new_sink_path, samplerate, _source.channels)
[4320679]59            # create a dictionary containing all this
[8b56b18]60            new_slice = {'start_stamp': start_stamp, 'end_stamp': end_stamp, 'sink': _sink}
[4320679]61            # append the dictionary to the current list of slices
62            slices.append(new_slice)
63
64        for current_slice in slices:
65            start_stamp = current_slice['start_stamp']
66            end_stamp = current_slice['end_stamp']
[8b56b18]67            _sink = current_slice['sink']
[4320679]68            # sample index to start writing from new source vector
69            start = max(start_stamp - total_frames, 0)
70            # number of samples yet to written be until end of region
71            remaining = end_stamp - total_frames + 1
72            #print current_slice, remaining, start
73            # not enough frames remaining, time to split
74            if remaining < read:
75                if remaining > start:
76                    # write remaining samples from current region
[8b56b18]77                    _sink.do_multi(vec[:, start:remaining], remaining - start)
[4320679]78                    #print "closing region", "remaining", remaining
79                    # close this file
[8b56b18]80                    _sink.close()
[4320679]81            elif read > start:
82                # write all the samples
[8b56b18]83                _sink.do_multi(vec[:, start:read], read - start)
[88432a9]84        total_frames += read
[04b31af]85        # remove old slices
86        slices = list(filter(lambda s: s['end_stamp'] > total_frames,
87            slices))
[8b56b18]88        if read < hopsize:
89            break
Note: See TracBrowser for help on using the repository browser.