def clean_tier(old_tier):
new_tier = IntervalTier(name=old_tier.name)
for interval in iter(old_tier):
new_tier.addInterval(clean_interval(interval))
return new_tier
def read_tg_from_str(tg_str, round_digits=DEFAULT_TEXTGRID_PRECISION):
"""
Read the tiers contained in the Praat-formatted string tg_str into a
TextGrid object.
Times are rounded to the specified precision.
Adapted from TextGrid.read()
"""
source = StringIO(tg_str)
tg = TextGrid()
file_type, short = parse_header(source)
if file_type != "TextGrid":
raise ValueError("The file could not be parsed as a TextGrid as it is "
"lacking a proper header.")
tg.minTime = parse_line(source.readline(), short, round_digits)
tg.maxTime = parse_line(source.readline(), short, round_digits)
source.readline() # More header junk
if short:
m = int(source.readline().strip()) # Will be tg.n
else:
m = int(source.readline().strip().split()[2]) # Will be tg.n
if not short:
source.readline()
for i in range(m): # Loop over grids
if not short:
source.readline()
if parse_line(source.readline(), short,
round_digits) == "IntervalTier":
inam = parse_line(source.readline(), short, round_digits)
imin = parse_line(source.readline(), short, round_digits)
imax = parse_line(source.readline(), short, round_digits)
itie = IntervalTier(inam, imin, imax)
itie.strict = tg.strict
n = int(parse_line(source.readline(), short, round_digits))
for j in range(n):
if not short:
source.readline().rstrip().split() # Header junk
jmin = parse_line(source.readline(), short, round_digits)
jmax = parse_line(source.readline(), short, round_digits)
jmrk = get_mark(source, short)
if jmin < jmax: # Non-null
itie.addInterval(Interval(jmin, jmax, jmrk))
tg.append(itie)
else: # PointTier
inam = parse_line(source.readline(), short, round_digits)
imin = parse_line(source.readline(), short, round_digits)
imax = parse_line(source.readline(), short, round_digits)
itie = PointTier(inam)
n = int(parse_line(source.readline(), short, round_digits))
for j in range(n):
source.readline().rstrip() # Header junk
jtim = parse_line(source.readline(), short, round_digits)
jmrk = get_mark(source, short)
itie.addPoint(Point(jtim, jmrk))
tg.append(itie)
return tg
def saveSyllableInTextGrid(sylIntervals, gridobj, f):
if len(sylIntervals) == 0: #No interval to save
return
minTime = sylIntervals[0].minTime #getting the min time for syllable tier
maxTime = sylIntervals[len(sylIntervals) -
1].maxTime #getting the max time for syllable tier
syllableTier = IntervalTier('Syllable', minTime, maxTime) #creating Tier
for interval in sylIntervals:
syllableTier.addInterval(interval)
gridobj.append(syllableTier) #appending Tier in text grid
gridobj.write(f) # writing the new Text Grid
def read(self, f):
"""
Read the tiers contained in the Praat-formated TextGrid file
indicated by string f
"""
source = readFile(f)
self.minTime = round(float(source.readline().split()[2]), 5)
self.maxTime = round(float(source.readline().split()[2]), 5)
source.readline() # more header junk
m = int(source.readline().rstrip().split()[2]) # will be self.n
source.readline()
for i in range(m): # loop over grids
source.readline()
if source.readline().rstrip().split()[2] == '"IntervalTier"':
inam = source.readline().rstrip().split(' = ')[1].strip('"')
inam = self.name_filter(inam)
imin = round(float(source.readline().rstrip().split()[2]), 5)
imax = round(float(source.readline().rstrip().split()[2]), 5)
itie = IntervalTier(inam)
for j in range(int(source.readline().rstrip().split()[3])):
source.readline().rstrip().split() # header junk
jmin = round(float(source.readline().rstrip().split()[2]),
5)
jmax = round(float(source.readline().rstrip().split()[2]),
5)
jmrk = _getMark(source)
if jmin < jmax: # non-null
itie.addInterval(Interval(jmin, jmax, jmrk))
self.append(itie)
else: # pointTier
inam = source.readline().rstrip().split(' = ')[1].strip('"')
inam = self.name_filter(inam)
imin = round(float(source.readline().rstrip().split()[2]), 5)
imax = round(float(source.readline().rstrip().split()[2]), 5)
itie = PointTier(inam)
n = int(source.readline().rstrip().split()[3])
for j in range(n):
source.readline().rstrip() # header junk
jtim = round(float(source.readline().rstrip().split()[2]),
5)
jmrk = source.readline().rstrip().split()[2][1:-1]
source.close()
def merge_and_mark_tiers(tg_file="", output_file="", tiers=()):
"""
Creates a new TextGrid file with an added IntervalTier.
"""
tg = textgrid.TextGrid()
tg.read(f=tg_file)
for t1_name, t2_name in combinations(tiers, 2):
validate_overlapping_tiers(tg.getFirst(t1_name), tg.getFirst(t2_name))
merged_tier = IntervalTier(
name="Merged",
minTime=min(map(lambda x: tg.getFirst(x).minTime, tiers)),
maxTime=min(map(lambda x: tg.getFirst(x).maxTime, tiers)))
marked_tier = IntervalTier(
name="Marked",
minTime=min(map(lambda x: tg.getFirst(x).minTime, tiers)),
maxTime=min(map(lambda x: tg.getFirst(x).maxTime, tiers)))
for tier_name, interval in filter(
lambda x: x[1].mark,
chain.from_iterable(
map(lambda x: zip(repeat(x.name), iter(x)),
map(lambda t: tg.getFirst(t), tiers)))):
marked_tier.addInterval(
Interval(minTime=interval.minTime,
maxTime=interval.maxTime,
mark=tier_name))
merged_tier.addInterval(
Interval(minTime=interval.minTime,
maxTime=interval.maxTime,
mark=interval.mark))
tg.tiers.insert(1, marked_tier)
tg.tiers.insert(2, merged_tier)
with open(output_file, "w") as f:
tg.write(f)
def read(self, f):
"""
Read the tiers contained in the Praat-formated TextGrid file
indicated by string f
"""
source = readFile(f)
self.minTime = round(float(source.readline().split()[2]), 5)
self.maxTime = round(float(source.readline().split()[2]), 5)
source.readline() # more header junk
m = int(source.readline().rstrip().split()[2]) # will be self.n
source.readline()
for i in range(m): # loop over grids
source.readline()
if source.readline().rstrip().split()[2] == '"IntervalTier"':
inam = source.readline().rstrip().split(' = ')[1].strip('"')
imin = round(float(source.readline().rstrip().split()[2]), 5)
imax = round(float(source.readline().rstrip().split()[2]), 5)
itie = IntervalTier(inam)
for j in range(int(source.readline().rstrip().split()[3])):
source.readline().rstrip().split() # header junk
jmin = round(float(source.readline().rstrip().split()[2]), 5)
jmax = round(float(source.readline().rstrip().split()[2]), 5)
jmrk = self._getMark(source)
if jmin < jmax: # non-null
itie.addInterval(Interval(jmin, jmax, jmrk))
self.append(itie)
else: # pointTier
inam = source.readline().rstrip().split(' = ')[1].strip('"')
imin = round(float(source.readline().rstrip().split()[2]), 5)
imax = round(float(source.readline().rstrip().split()[2]), 5)
itie = PointTier(inam)
n = int(source.readline().rstrip().split()[3])
for j in range(n):
source.readline().rstrip() # header junk
jtim = round(float(source.readline().rstrip().split()[2]),
5)
jmrk = source.readline().rstrip().split()[2][1:-1]
source.close()
if i == 0: for x in ti: x.maxTime += cur_dur x.minTime += cur_dur wordintervals.append(x) elif i == 1: for x in ti: x.maxTime += cur_dur x.minTime += cur_dur phoneintervals.append(x) cur_dur += maxtime words = IntervalTier(name='words') for i in wordintervals: words.addInterval(i) phones = IntervalTier(name='phones') for i in phoneintervals: phones.addInterval(i) tg1 = TextGrid(maxTime = cur_dur) tg1.append(words) tg1.append(phones) tg1.write(chapteroutpath1, null = '') speaker_tier = IntervalTier(name=speaker) for i in range(len(groupedwavfiles)): if i == 1: speaker_tier.add(0.0, wavfiletimes[0], groupedlabtext[0]) else: speaker_tier.add(wavfiletimes[i-2], wavfiletimes[i-1], groupedlabtext[i-1]) tg2 = TextGrid(maxTime = duration)
if i == 0:
for x in ti:
x.maxTime += cur_dur
x.minTime += cur_dur
wordintervals.append(x)
elif i == 1:
for x in ti:
x.maxTime += cur_dur
x.minTime += cur_dur
phoneintervals.append(x)
cur_dur += maxtime
words = IntervalTier(name='words')
for i in wordintervals:
words.addInterval(i)
phones = IntervalTier(name='phones')
for i in phoneintervals:
phones.addInterval(i)
tg1 = TextGrid(maxTime=cur_dur)
tg1.append(words)
tg1.append(phones)
tg1.write(chapteroutpath1, null='')
speaker_tier = IntervalTier(name=speaker)
for i in range(len(groupedwavfiles)):
if i == 1:
speaker_tier.add(0.0, wavfiletimes[0], groupedlabtext[0])
else:
speaker_tier.add(wavfiletimes[i - 2], wavfiletimes[i - 1],
groupedlabtext[i - 1])
for s in sorted(speaker_word_tiers.keys()):
w_tier = IntervalTier('{} - word'.format(s), 0, duration)
p_tier = IntervalTier('{} - phone'.format(s), 0, duration)
for w in sorted(speaker_word_tiers[s]):
if len(w_tier) and w_tier[-1].mark in [
'sp', '{OOV}'
] and w_tier[-1].maxTime > w.minTime:
w_tier[-1].maxTime = w.minTime
if len(w_tier) and w.mark in [
'sp', '{OOV}'
] and w_tier[-1].maxTime > w.minTime:
w.minTime = w_tier[-1].maxTime
#print(w)
if w.maxTime > duration:
w.maxTime = duration
w_tier.addInterval(w)
for p in sorted(speaker_phone_tiers[s]):
if len(p_tier) and p_tier[
-1].mark == 'sil' and p_tier[-1].maxTime > p.minTime:
p_tier[-1].maxTime = p.minTime
if len(p_tier) and p.mark == 'sil' and p_tier[
-1].maxTime > p.minTime:
p.minTime = p_tier[-1].maxTime
#print(p)
if p.maxTime > duration:
p.maxTime = duration
try:
p_tier.addInterval(p)
except ValueError:
pass
new_tg.append(w_tier)
