def run_ipusegmentation(self, stepidx):
"""
Execute the SPPAS-IPUSegmentation program.
@return number of files processed successfully
"""
# Initializations
step = self.parameters.get_step(stepidx)
stepname = self.parameters.get_step_name(stepidx)
files_processed_success = 0
self._progress.set_header(stepname)
self._progress.update(0,"")
# Get the list of input file names, with the ".wav" extension
filelist = self.set_filelist(".wav")
if len(filelist) == 0:
return 0
total = len(filelist)
# Create annotation instance, and fix options
try:
seg = sppasSeg(self._logfile)
except Exception as e:
if self._logfile is not None:
self._logfile.print_message( "%s\n"%str(e), indent=1,status=1 )
return 0
# Execute the annotation for each file in the list
for i,f in enumerate(filelist):
# fix the default values
seg.fix_options( step.get_options() )
# Indicate the file to be processed
if self._logfile is not None:
self._logfile.print_message(stepname+" of file "+f, indent=1 )
self._progress.set_text( os.path.basename(f)+" ("+str(i+1)+"/"+str(total)+")" )
# Fix input/output file name
outname = os.path.splitext(f)[0] + self.parameters.get_output_format()
# Is there already an existing IPU-seg (in any format)!
existoutname = self._get_filename(f, ['.xra', '.TextGrid', '.eaf', '.csv', '.mrk' ])
# it's existing... but not in the expected format: convert!
if existoutname is not None and existoutname != outname:
# just copy the file!
if self._logfile is not None:
self._logfile.print_message('Export '+existoutname, indent=2)
self._logfile.print_message('into '+outname, indent=2)
try:
t = annotationdata.io.read(existoutname)
annotationdata.io.write(outname,t)
# OK, now outname is as expected! (or not...)
except Exception:
pass
# Execute annotation
tgfname = utils.fileutils.exists(outname)
if tgfname is None:
# No already existing IPU seg., but perhaps a txt.
txtfile = self._get_filename(f, [".txt"])
if self._logfile is not None:
if txtfile:
self._logfile.print_message("A transcription was found, perform Silence/Speech segmentation time-aligned with a transcription %s"%txtfile, indent=2,status=3)
else:
self._logfile.print_message("No transcription was found, perform Silence/Speech segmentation only.", indent=2,status=3)
try:
seg.run(f, trsinputfile=txtfile, ntracks=None, diroutput=None, tracksext=None, listoutput=None, textgridoutput=outname)
files_processed_success += 1
if self._logfile is not None:
self._logfile.print_message(outname, indent=2,status=0)
except Exception as e:
if self._logfile is not None:
self._logfile.print_message( "%s for file %s\n"%(str(e),outname), indent=2,status=-1 )
else:
if seg.get_dirtracks() is True:
self._logfile.print_message("A time-aligned transcription was found, split into multiple files", indent=2,status=3)
try:
seg.run(f, trsinputfile=tgfname, ntracks=None, diroutput=None, tracksext=None, listoutput=None, textgridoutput=None)
files_processed_success += 1
if self._logfile is not None:
self._logfile.print_message(tgfname, indent=2,status=0 )
except Exception as e:
if self._logfile is not None:
self._logfile.print_message( "%s for file %s\n"%(str(e),tgfname), indent=2,status=-1 )
else:
if self._logfile is not None:
self._logfile.print_message( "because a previous segmentation is existing.", indent=2,status=2 )
# Indicate progress
self._progress.set_fraction(float((i+1))/float(total))
if self._logfile is not None:
self._logfile.print_newline()
# Indicate completed!
self._progress.update(1,"Completed (%d files successfully over %d files).\n"%(files_processed_success,total))
self._progress.set_header("")
return files_processed_success
sys.exit(1)
if args.t and args.e and args.t.lower().endswith('.txt'):
print "You can NOT use both -e with -t option if file is not time-aligned!\nUse option -h for any help."
sys.exit(1)
# ----------------------------------------------------------------------------
# Automatic IPUs segmentation is here:
# ----------------------------------------------------------------------------
if args.o and not os.path.exists(args.o):
os.mkdir(args.o)
w = sppasSeg()
if args.shift: w.set_shift( args.shift )
if args.minvol: w.set_min_volume( args.minvol )
if args.minsil: w.set_min_silence( args.minsil )
if args.minrms: w.set_min_speech( args.minrms )
if args.o: w.set_dirtracks( True )
if args.p: w.set_save_as_trs( True )
if args.e: w.set_save_as_trs( True )
w.run(args.w, args.t, args.n, args.N, args.o, args.e, args.l, args.p)
# - wavfile is the wav input file name
# - trsinputfile is a transcription (or 'None')
# - trstieridx is the tier index with the transcription
# - ntracks