def read(xml, max_time=4, max_words=None):
"""
Transform XML, an ElementTree.Element representation of Koemei's
<segmentation> output, into a series of spindle clips and associated
speaker objects.
Returns a dict with two values: CLIPS and SPEAKERS
"""
speakers = dict()
clips = []
for segment in xml.findall('segment'):
start = float(segment.find('start').text.strip()) / 100
end = float(segment.find('end').text.strip()) / 100
speaker_name = segment.find('speaker').text.strip()
try:
speaker = speakers[speaker_name]
except KeyError:
speaker = Speaker(name=speaker_name)
speakers[speaker_name] = speaker
segment_clips = words_to_clips(segment_to_words(segment),
max_time=max_time,
max_words=max_words,
speaker=speaker)
clip = segment_clips.next()
clip.begin_para = True
clips.append(clip)
for clip in segment_clips:
clips.append(clip)
return dict(clips=clips, speakers=speakers.values())
def read(xml, max_time=4, max_words=None):
"""
Transform XML, an ElementTree.Element representation of Koemei's
<segmentation> output, into a series of spindle clips and associated
speaker objects.
Returns a dict with two values: CLIPS and SPEAKERS
"""
speakers = dict()
clips = []
for segment in xml.findall('segment'):
start = float(segment.find('start').text.strip()) / 100
end = float(segment.find('end').text.strip()) / 100
speaker_name = segment.find('speaker').text.strip()
try:
speaker = speakers[speaker_name]
except KeyError:
speaker = Speaker(name=speaker_name)
speakers[speaker_name] = speaker
segment_clips = words_to_clips(segment_to_words(segment),
max_time=max_time,
max_words=max_words,
speaker=speaker)
clip = segment_clips.next()
clip.begin_para = True
clips.append(clip)
for clip in segment_clips: clips.append(clip)
return dict(clips=clips, speakers=speakers.values())
def read_clips(input, **kwargs):
"""Read Sphinx4 output and segment into clips.
`input' should be a file object, or anything that can be iterated
to produce lines of text.
Any keyword arguments are passed unchanged to words_to_clips.
Yields Clip objects segmented in two ways:
(1) Sphinx segment tokens <s> begin a new clip
(2) Within Sphinx segments, words are split into clips using
words_to_clips, so by default each clip is limited to a maximum
length of 4 seconds."""
for segment in segments(remove_silences(read_tokens(input))):
for clip in words_to_clips(segment, **kwargs):
yield clip
def read_clips(input, **kwargs):
"""Read Sphinx4 output and segment into clips.
`input' should be a file object, or anything that can be iterated
to produce lines of text.
Any keyword arguments are passed unchanged to words_to_clips.
Yields Clip objects segmented in two ways:
(1) Sphinx segment tokens <s> begin a new clip
(2) Within Sphinx segments, words are split into clips using
words_to_clips, so by default each clip is limited to a maximum
length of 4 seconds."""
for segment in segments(remove_silences(read_tokens(input))):
for clip in words_to_clips(segment, **kwargs):
yield clip