def split_speech_eaf(in_folder, out_folder, num_threads):
file_finder = helper.FindFiles("file_finder", dir=in_folder, ext=".wav")
splitter = EafSegments(in_folder)
split = audio.SplitSegments("split_speech", out_dir=out_folder, segment_mapping_fn=splitter.get_segs_for_file)
p = ProgressPipeline(file_finder | split, n_threads=num_threads, quiet=True)
return p
def main(in_folder, out_folder, num_threads):
file_finder = helper.FindFiles("file_finder", dir=in_folder, ext=".wav")
resample = audio.ResampleWav("resampled_audio",
out_dir=out_folder,
new_sr=8000)
is10_lld = audio.OpenSmileRunner("is10_lld",
out_dir=out_folder,
conf_file="IS10_paraling.conf",
out_flag="-lldcsvoutput")
is10 = audio.OpenSmileRunner("is10",
out_dir=out_folder,
conf_file="IS10_paraling.conf",
out_flag="-csvoutput")
splitter = EafSegments(in_folder)
split = audio.SplitSegments("utterances",
out_dir=out_folder,
segment_mapping_fn=splitter.get_segs_for_file,
out_streams=["audio", "label"])
is10_lld_per_utterance = audio.OpenSmileRunner(
"is10_lld_per_utterance",
out_dir=out_folder,
in_streams="audio",
conf_file="IS10_paraling.conf",
out_flag="-lldcsvoutput")
is10_per_utterance = audio.OpenSmileRunner("is10_per_utterance",
out_dir=out_folder,
in_streams="audio",
conf_file="IS10_paraling.conf",
out_flag="-csvoutput")
asr = audio.KaldiASR("kaldi_asr", out_dir=out_folder, in_streams="audio")
p = ProgressPipeline(file_finder | resample | [
is10_lld, is10,
split | [asr, is10_lld_per_utterance, is10_per_utterance]
],
n_threads=num_threads,
quiet=True)
return p