def compute_gamma(self, ref_tg: TextGrid,
target_tg: TextGrid) -> Optional[float]:
continuum = Continuum()
for annot in ref_tg.getFirst(self.name):
continuum.add("ref", Segment(annot.minTime, annot.maxTime))
for annot in target_tg.getFirst(self.name):
continuum.add("target", Segment(annot.minTime, annot.maxTime))
dissim = PositionalSporadicDissimilarity(delta_empty=1)
gamma_results = continuum.compute_gamma(dissim,
n_samples=10,
precision_level="medium")
return gamma_results.gamma
def compute_gamma(self, ref_tg: TextGrid, target_tg: TextGrid):
continuum = Continuum()
for annot in ref_tg.getFirst(self.name):
continuum.add("ref", Segment(annot.minTime, annot.maxTime),
annot.mark)
for annot in target_tg.getFirst(self.name):
continuum.add("target", Segment(annot.minTime, annot.maxTime),
annot.mark)
dissim = CombinedCategoricalDissimilarity(alpha=1, beta=1)
gamma_results = continuum.compute_gamma(dissim,
n_samples=10,
precision_level="medium")
return gamma_results.gamma
def reorg_original(f):
print(f)
tg_path = os.path.join(noncollapsed_dir, f)
tg = TextGrid()
tg.read(tg_path)
new_tg = TextGrid(maxTime=tg.maxTime)
new_tg_path = tg_path.replace('_original.TextGrid', '.TextGrid')
sentence_tier = tg.getFirst('Sentences')
speaker_tier = tg.getFirst('Speakers')
speaker_tiers = {}
for i in speaker_tier:
if i.mark == '':
continue
if ',' in i.mark:
continue
if i.mark == 'Tân.':
continue
speaker_tiers[i.mark] = IntervalTier(i.mark, maxTime=tg.maxTime)
for i in sentence_tier:
if not i.mark.strip():
continue
duration = i.maxTime - i.minTime
mid_point = i.minTime + duration / 2
speaker_int = speaker_tier.intervalContaining(mid_point)
speaker = speaker_int.mark
if speaker == 'Tân.':
speaker = 'Tan'
if speaker == '':
continue
if len(speaker_tiers[speaker]
) > 0 and speaker_tiers[speaker][-1].maxTime == i.minTime:
speaker_tiers[speaker][-1].maxTime = i.maxTime
speaker_tiers[speaker][
-1].mark = speaker_tiers[speaker][-1].mark + ' ' + i.mark
else:
speaker_tiers[speaker].addInterval(i)
for k, v in sorted(speaker_tiers.items()):
new_tg.append(v)
print(speaker_tiers.keys())
new_tg.write(new_tg_path)
def _load_tier(
grid: textgrid.TextGrid,
tier: str = 'phones',
clean: bool = True,
):
"""Load one or more tiers as textgrid Tier object"""
names = [name for name in grid.getNames() if fnmatch.fnmatch(name.lower(), tier.lower())]
if len(names) != 1:
available = ', '.join(grid.getNames())
raise IOError(f"{len(names)} tiers match {tier!r} in {grid.name or grid}. Availabe tiers: {available}")
tier = grid.getFirst(names[0])
if clean:
for item in tier:
item.mark = item.mark.strip()
if item.mark in SILENCE:
item.mark = ' '
return tier
def ctm_to_textgrid(word_ctm, phone_ctm, out_directory, corpus):
if not os.path.exists(out_directory):
os.makedirs(out_directory, exist_ok=True)
if not corpus.segments:
for i,(k,v) in enumerate(word_ctm.items()):
maxtime = corpus.get_wav_duration(k)
try:
tg = TextGrid(maxTime = maxtime)
wordtier = IntervalTier(name = 'words', maxTime = maxtime)
phonetier = IntervalTier(name = 'phones', maxTime = maxtime)
for interval in v:
wordtier.add(*interval)
for interval in phone_ctm[k]:
phonetier.add(*interval)
tg.append(wordtier)
tg.append(phonetier)
if corpus.speaker_directories:
speaker_directory = os.path.join(out_directory, corpus.utt_speak_mapping[k])
else:
speaker_directory = out_directory
os.makedirs(speaker_directory, exist_ok=True)
outpath = os.path.join(speaker_directory, k + '.TextGrid')
tg.write(outpath)
except ValueError as e:
print('Could not write textgrid for {}'.format(k))
print(e)
else:
tgs = {}
for i,(k,v) in enumerate(word_ctm.items()):
rec = corpus.segments[k]
rec, begin, end = rec.split(' ')
maxtime = corpus.get_wav_duration(k)
if rec not in tgs:
tgs[rec] = TextGrid(maxTime = maxtime)
tg = tgs[rec]
begin = float(begin)
speaker = corpus.utt_speak_mapping[k]
word_tier_name = '{} - words'.format(speaker)
phone_tier_name = '{} - phones'.format(speaker)
wordtier = tg.getFirst(word_tier_name)
if wordtier is None:
wordtier = IntervalTier(name = word_tier_name, maxTime = maxtime)
tg.append(wordtier)
phonetier = tg.getFirst(phone_tier_name)
if phonetier is None:
phonetier = IntervalTier(name = phone_tier_name, maxTime = maxtime)
tg.append(phonetier)
for interval in v:
interval = interval[0] + begin, interval[1] + begin, interval[2]
wordtier.add(*interval)
for interval in phone_ctm[k]:
interval = interval[0] + begin, interval[1] + begin, interval[2]
phonetier.add(*interval)
for k,v in tgs.items():
outpath = os.path.join(out_directory, k + '.TextGrid')
try:
v.write(outpath)
except ValueError as e:
print('Could not write textgrid for {}'.format(k))
print(e)
def parse_dataset(
directory: Path,
tier_name: str = DEFAULT_TIER_NAME,
n_digits: int = DEFAULT_N_DIGITS,
encoding: str = DEFAULT_ENCODING,
audio_format: str = DEFAULT_AUDIO_FORMAT
) -> Generator[Entry, None, None]:
if not directory.is_dir():
raise ValueError("Parameter 'directory': Directory was not found!")
if not isinstance(tier_name, str):
raise ValueError(
"Parameter 'tier_name: Value needs to be of type 'str'!")
if n_digits not in range(1, 17):
raise ValueError(
"Parameter 'n_digits': Value needs to be in interval [0, 16]!")
if not isinstance(encoding, str):
raise ValueError(
"Parameter 'encoding': Value needs to be of type 'str'!")
logger = getLogger(__name__)
speaker_dirs = get_subfolders(directory)
for speaker_dir in speaker_dirs:
speaker_parts = speaker_dir.name.split(PARTS_SEP)
if len(speaker_parts) not in {3, 4}:
logger.warning(
f"{str(speaker_dir.relative_to(directory))}: Directory '{speaker_dir.name}' couldn't be parsed because not all information are provided in the name. Ignored."
)
continue
speaker_name = speaker_parts[0]
speaker_gender = speaker_parts[1]
if not speaker_gender.isnumeric():
logger.warning(
f"{str(speaker_dir.relative_to(directory))}: Gender code '{speaker_gender}' needs to be a number. Ignored."
)
continue
speaker_gender = int(speaker_gender)
if not speaker_gender in GENDERS:
logger.warning(
f"{str(speaker_dir.relative_to(directory))}: Gender code '{speaker_gender}' not recognized. Ignored."
)
continue
speaker_lang = speaker_parts[2]
# TODO check lang code better
if len(speaker_lang) != 3 or not speaker_lang.islower():
logger.warning(
f"{str(speaker_dir.relative_to(directory))}: Language code '{speaker_lang}' is not valid (needs to be three lower-case letters). Ignored."
)
continue
speaker_accent = None
if len(speaker_parts) == 4:
speaker_accent = speaker_parts[3]
audio_files = get_files_dict(speaker_dir, {audio_format})
grid_files = get_files_dict(speaker_dir, {".TextGrid"})
for file_stem, grid_file_rel in tqdm(grid_files.items()):
if file_stem not in audio_files:
logger.warning(
f"{str(grid_file_rel)}: Audio file was not found. Ignored."
)
continue
grid_file_abs = speaker_dir / grid_file_rel
grid = TextGrid()
grid.read(grid_file_abs, n_digits, encoding)
tier = cast(Optional[IntervalTier], grid.getFirst(tier_name))
if tier is None:
logger.warning(
f"{str(grid_file_rel)}: Tier '{tier_name}' does not exist! Ignored."
)
symbols = (
interval.mark
for interval in cast(Iterable[Interval], tier.intervals))
symbols = tuple(symbol if symbol is not None else ""
for symbol in symbols)
intervals = tuple(
interval.maxTime
for interval in cast(Iterable[Interval], tier.intervals))
assert len(symbols) == len(intervals)
audio_path = speaker_dir / audio_files[file_stem]
result = Entry(symbols, intervals, speaker_lang, speaker_name,
speaker_accent, speaker_gender, audio_path,
grid.minTime, grid.maxTime)
yield result
from textgrid import TextGrid
import json
tg = TextGrid()
tg.read("./temp.TextGrid")
data = []
for i in tg.getFirst("words"):
data.append({
"begin": str(i.minTime),
"end": str(i.maxTime),
"word": i.mark
})
with open('out.json', 'w') as outfile:
json.dump(data, outfile)
class TextGrid(Module):
'''
Manages all the widgets related to TextGrid files, including the tier name
and the text content of that tier at a given frame
'''
def __init__(self, app):
'''
Keep a reference to the master object for binding the widgets we create
'''
info(' - initializing module: TextGrid')
self.app = app
self.frame = Frame(self.app.BOTTOM)
self.label_padx = 0
self.canvas_frame = Frame(self.app.BOTTOM) #, padx=self.label_padx)
self.frame.grid(row=1, column=0, sticky='ne')
self.canvas_frame.grid(row=1, column=1)
self.TextGrid = None
self.selectedTier = StringVar()
self.tg_zoom_factor = 1.5
self.canvas_width = 800
self.canvas_height = 60
self.collapse_height = 15
self.selectedIntvlFrames = []
self.selectedItem = None
self.start = 0
self.end = 0
self.current = 0
self.frame_shift = DoubleVar()
self.startup()
platform = util.get_platform()
#bindings
if platform == 'Linux':
self.app.bind("<Control-n>", self.getBounds)
self.app.bind("<Control-a>", self.getBounds)
self.app.bind("<Control-i>", self.getBounds)
self.app.bind("<Control-o>", self.getBounds)
self.app.bind("<Control-f>", self.openSearch)
# Command is Alt in Linux, apparently
self.app.bind("<Command-Up>", self.changeTiers)
self.app.bind("<Command-Down>", self.changeTiers)
self.app.bind("<Command-Left>", self.changeIntervals)
self.app.bind("<Command-Right>", self.changeIntervals)
elif platform == 'Darwin':
self.app.bind("<Command-n>", self.getBounds)
self.app.bind("<Command-a>", self.getBounds)
self.app.bind("<Command-i>", self.getBounds)
self.app.bind("<Command-o>", self.getBounds)
self.app.bind("<Command-f>", self.openSearch)
self.app.bind("<Option-Up>", self.changeTiers)
self.app.bind("<Option-Down>", self.changeTiers)
self.app.bind("<Option-Left>", self.changeIntervals)
self.app.bind("<Option-Right>", self.changeIntervals)
#defaults (Command/Alt everything)
else:
self.app.bind("<Command-n>", self.getBounds)
self.app.bind("<Command-a>", self.getBounds)
self.app.bind("<Command-i>", self.getBounds)
self.app.bind("<Command-o>", self.getBounds)
self.app.bind("<Command-f>", self.openSearch)
self.app.bind("<Command-Up>", self.changeTiers)
self.app.bind("<Command-Down>", self.changeTiers)
self.app.bind("<Command-Left>", self.changeIntervals)
self.app.bind("<Command-Right>", self.changeIntervals)
# these aren't Praat-like
self.app.bind("<Shift-Left>", self.getBounds)
self.app.bind("<Shift-Right>", self.getBounds)
def setup(self):
if LIBS_INSTALLED:
self.loadOrGenerate()
try:
self.start = self.TextGrid.minTime
self.end = self.TextGrid.maxTime
self.app.frames = len(
self.TextGrid.getFirst(self.frameTierName))
tiers = []
for tier in self.TextGrid.getNames():
if tier != self.frameTierName and tier != self.frameTierName + '.original':
tiers.append(tier)
if set(tiers) != self.tierNames:
self.tierNames = set(tiers)
self.TkWidgets = []
for label in self.frame.winfo_children():
label.destroy()
for canvas in self.canvas_frame.winfo_children():
canvas.destroy()
for tier in self.TextGrid.getNames():
if tier != self.frameTierName and tier != self.frameTierName + '.original':
self.TkWidgets.append(self.makeTierWidgets(tier))
self.makeFrameWidget()
self.makeTimeWidget()
self.fillCanvases()
self.firstFrame = int(
self.TextGrid.getFirst(self.frameTierName)[0].mark) + 1
self.startFrame = self.firstFrame
self.lastFrame = int(
self.TextGrid.getFirst(self.frameTierName)[-1].mark) + 1
self.endFrame = self.lastFrame
self.grid()
except Exception as e:
error(e)
def startup(self):
'''
'''
if LIBS_INSTALLED:
self.tierNames = set()
self.setup()
def reset(self, event=None):
'''
Try to load a TextGrid file based on information stored in the metadata
'''
if LIBS_INSTALLED:
self.selectedIntvlFrames = []
self.selectedItem = None
self.setup()
def fromFile(self, filename):
if LIBS_INSTALLED:
try:
return TextGridFile.fromFile(filename)
except (TextGridError, UnicodeDecodeError) as e:
f = open(filename, 'rb')
contents = util.decode_bytes(f.read())
f.close()
if contents:
tmp = tempfile.NamedTemporaryFile()
tmp.write(contents.encode('utf-8'))
tmp.seek(0)
try:
return TextGridFile.fromFile(tmp.name)
except TextGridError as e:
error(e)
return None
else:
error(
"can't load from file: unable to decode non-Unicode textgrid",
filename)
else:
error("can't load from file: textgrid lib not installed")
return None
def loadOrGenerate(self):
fname = self.app.Data.checkFileLevel('.TextGrid', shoulderror=False)
if fname:
self.TextGrid = self.fromFile(fname)
else:
minTime = 0.
if not hasattr(self.app.Audio, 'duration'):
self.app.Audio.reset()
try:
maxTime = self.app.Audio.duration
except:
warn(
'Audio has no duration attribute after calling reset(), defaulting to 1 second'
)
maxTime = 1.
self.TextGrid = TextGridFile(maxTime=maxTime)
keys = self.app.Data.getFileLevel('all')
if not ('.ult' in keys and '.txt' in keys):
sentenceTier = IntervalTier("text")
sentenceTier.add(minTime, maxTime, "text")
self.TextGrid.append(sentenceTier)
fname = self.app.Data.unrelativize(
self.app.Data.getCurrentFilename() + '.TextGrid')
self.app.Data.setFileLevel('.TextGrid', fname)
names = self.TextGrid.getNames()
for i, n in enumerate(names):
if n in ALIGNMENT_TIER_NAMES:
if len(self.TextGrid[i]) == 0:
self.TextGrid.pop(i)
break
else:
self.frameTierName = n
return
self.genFramesTier()
def genFramesTier(self):
debug('generating frames tier for %s' %
self.app.Data.getCurrentFilename())
self.frameTierName = 'frames'
times = self.app.Dicom.getFrameTimes()
self.app.Data.setFileLevel("NumberOfFrames", len(times))
try:
maxTime = max(self.app.Audio.duration, times[-1])
except AttributeError:
maxTime = times[-1]
tier = PointTier('frames', maxTime=maxTime)
for f, t in enumerate(times):
tier.addPoint(Point(t, str(f)))
if not self.TextGrid.maxTime or maxTime > self.TextGrid.maxTime:
self.TextGrid.maxTime = maxTime
self.TextGrid.append(tier)
keys = self.app.Data.getFileLevel('all')
if '.ult' in keys and '.txt' in keys:
fname = self.app.Data.unrelativize(
self.app.Data.getFileLevel('.txt'))
f = open(fname, 'rb')
s = util.decode_bytes(f.read())
f.close()
if s:
line = s.splitlines()[0]
sentenceTier = IntervalTier("sentence")
sentenceTier.add(0, self.app.Audio.duration, line)
self.TextGrid.append(sentenceTier)
self.TextGrid.tiers = [self.TextGrid.tiers[-1]
] + self.TextGrid.tiers[:-1]
path = self.app.Data.unrelativize(
self.app.Data.getFileLevel('.TextGrid'))
self.TextGrid.write(path)
self.TextGrid = TextGridFile.fromFile(path)
# reload to account for length changes due to frames tier being different length than audio
@staticmethod
def isIntervalTier(tier):
if LIBS_INSTALLED:
return isinstance(tier, IntervalTier)
else:
error("can't check if IntervalTier: textgrid lib not installed")
return False
def shiftFrames(self):
'''
Replicate original TextGrid point tier (renamed [tiername].original)
Shift points on TextGrid tier in accordance with self.frame_shift
Shift value is relative to 0, i.e. inputting the same shift amount a second time will not change the shift
Redisplay shifted points
'''
self.app.focus()
shift = self.frame_shift.get()
if type(shift) == float:
self.app.Data.setFileLevel('offset', shift)
# diff = shift - self.app.Data.data['offset']
originalTier = self.TextGrid.getFirst(self.frameTierName +
'.original')
if originalTier: pass
else:
orig = copy.deepcopy(self.TextGrid.getFirst(
self.frameTierName))
orig.name += '.original'
self.TextGrid.append(orig)
originalTier = self.TextGrid.getFirst(self.frameTierName +
'.original')
oldTier = self.TextGrid.getFirst(self.frameTierName)
allPoints = oldTier[:]
for point in allPoints:
oldTier.removePoint(point)
for point in originalTier:
new_time = point.time + shift / 1000 ## NOTE: currently in ms
if self.TextGrid.minTime <= new_time <= self.TextGrid.maxTime:
self.TextGrid.getFirst(self.frameTierName).add(
new_time, point.mark)
# self.app.frames = len(self.TextGrid.getFirst(self.frameTierName)) #FIXME I feel like I shouldn't have to run the getFirst function every time, but I'm not sure when I have to go back to the original textgrid, and when I can just use a variable...
self.firstFrame = int(
self.TextGrid.getFirst(self.frameTierName)[0].mark) + 1
self.lastFrame = int(
self.TextGrid.getFirst(self.frameTierName)[-1].mark) + 1
self.app.Data.data['offset'] = shift
# self.frame_shift.set(shift)
self.app.Data.write()
# newTier.write(self.TextGrid.getFirst(self.frameTierName))
self.fillCanvases()
self.TextGrid.write(
self.app.Data.unrelativize(
self.app.Data.getFileLevel('.TextGrid')))
#except ValueError:
else:
error('Not a float!')
def makeTimeWidget(self):
self.time_canvas = Canvas(self.canvas_frame,
width=self.canvas_width,
height=self.canvas_height / 3,
highlightthickness=0)
s = self.time_canvas.create_text(3, 0, anchor='nw', text=self.start)
e = self.time_canvas.create_text(self.canvas_width,
0,
anchor='ne',
text=self.end)
c = self.time_canvas.create_text(self.canvas_width / 2,
0,
anchor='n',
text=self.current)
self.TkWidgets.append({'times': self.time_canvas})
def makeFrameWidget(self):
'''
makes frame widget
'''
#make regular frame stuff -- label and tier
self.frames_canvas = Canvas(self.canvas_frame,
width=self.canvas_width,
height=self.canvas_height,
background='gray',
highlightthickness=0)
frames_label = Canvas(self.frame,
width=self.label_width,
height=self.canvas_height,
highlightthickness=0,
background='gray')
frames_label.create_text(self.label_width,
0,
anchor='ne',
justify='center',
text='frames: ',
width=self.label_width,
activefill='blue')
# make subframe to go on top of label canvas
sbframe = Frame(frames_label)
#put new widgets onto subframe
offset = self.app.Data.getFileLevel('offset')
if offset != None:
self.frame_shift.set(offset)
# for audio alignment
go_btn = Button(sbframe,
text='Offset',
command=self.shiftFrames,
takefocus=0)
# minmax = len(self.app.Audio.sfile)*1000
txtbox = Spinbox(sbframe,
textvariable=self.frame_shift,
width=7,
from_=-10000000,
to=10000000)
txtbox.bind('<Escape>', lambda ev: sbframe.focus())
txtbox.bind('<Return>', lambda ev: self.shiftFrames())
go_btn.grid(row=0, column=0, sticky='e')
txtbox.grid(row=0, column=1, sticky='e')
# put subframe on canvas
window = frames_label.create_window(self.label_width * .3,
self.canvas_height / 3,
anchor='nw',
window=sbframe)
# ensure position of subframe gets updated
frames_label.bind(
'<Configure>',
lambda e: frames_label.itemconfig(window, width=e.width))
sbframe.bind(
'<Configure>', lambda e: frames_label.configure(
scrollregion=frames_label.bbox("all")))
self.TkWidgets.append({
'name': self.frameTierName,
'frames': self.frames_canvas,
'frames-label': frames_label
})
self.frames_canvas.bind("<Button-1>", self.getClickedFrame)
def getFrameTierName(self):
'''
Handle some inconsistency in how we're naming our alignment tier
'''
for name in ALIGNMENT_TIER_NAMES:
if name in self.TextGrid.getNames():
return name
raise NameError('Unable to find alignment tier')
def getClickedFrame(self, event):
'''
Jumps to clicked frame
'''
item = self.my_find_closest(event.widget, event.x)
self.setSelectedIntvlFrames((event.widget, item))
frame = event.widget.gettags(item)[0][5:]
self.app.frame = int(frame)
if not frame in self.selectedIntvlFrames:
self.selectedIntvlFrames = []
self.wipeFill()
self.app.framesUpdate()
def makeTierWidgets(self, tier):
'''
Each tier should have two canvas widgets: `canvas-label` (the tier name),
and `canvas` (the intervals on the tier with their marks)
'''
self.tier_pairs = {} #ends up being format {label: canvas}
# self.app.Trace.frame.update()
self.label_width = 300 #self.app.Trace.frame.winfo_width()+self.label_padx
self.end = self.TextGrid.maxTime #float(self.TextGrid.maxTime)
# self.first_frame = 1
# self.last_frame = self.TextGrid.getFirst(self.frameTierName)[-1].mark
tier_obj = self.TextGrid.getFirst(tier)
widgets = {
'name':
tier,
#'label':Label(self.frame, text=('- '+tier+':'), wraplength=200, justify='left'),
'canvas-label':
Canvas(self.frame,
width=self.label_width,
height=self.canvas_height,
highlightthickness=0),
# 'text' :Label(self.frame, text='', wraplength=550, justify='left'),
'canvas':
Canvas(self.canvas_frame,
width=self.canvas_width,
height=self.canvas_height,
background='gray',
highlightthickness=0)
}
canvas = widgets['canvas']
label = widgets['canvas-label']
#builds tier label functionality
label_text = label.create_text(self.label_width,
self.canvas_height / 2,
anchor='e',
justify='center',
text='temp',
width=self.label_width / 2,
activefill='blue')
canvas.bind("<Button-1>", self.genFrameList)
label.bind("<Button-1>", self.genFrameList)
label.bind("<Double-Button-1>", self.collapse)
label.bind("<Button-4>", self.collapse)
label.bind("<Button-5>", self.collapse)
label.bind("<MouseWheel>", self.collapse)
canvas.bind("<Button-4>", self.collapse)
canvas.bind("<Button-5>", self.collapse)
canvas.bind("<MouseWheel>", self.collapse)
return widgets
def changeIntervals(self, event):
'''
'''
if self.selectedItem:
duration = self.end - self.start
# There might be a more efficient way to get the tier name:
widg = self.selectedItem[0]
itm = self.selectedItem[1]
for el in self.TkWidgets:
if widg in el.values():
tier_name = el['name']
break
#finding Interval mintime and maxtime
oldMinTime = None
oldMaxTime = None
q = 0
tags = widg.gettags(itm)
while oldMinTime == None or oldMaxTime == None:
if tags[q][:7] == 'minTime':
oldMinTime = float(tags[q][7:])
elif tags[q][:7] == 'maxTime':
oldMaxTime = float(tags[q][7:])
q += 1
tier = self.TextGrid.getFirst(tier_name)
intvl_i = tier.indexContaining(oldMaxTime -
((oldMaxTime - oldMinTime) / 2))
if event.keysym == 'Left':
new_intvl_i = intvl_i - 1
elif event.keysym == 'Right':
new_intvl_i = intvl_i + 1
if 0 <= new_intvl_i < len(tier):
#find characteristics of new adjacent interval
newMinTime = tier[new_intvl_i].minTime
newMaxTime = tier[new_intvl_i].maxTime
itvlDuration = newMaxTime - newMinTime
newCenter = newMinTime + itvlDuration / 2
#figure out new window parameters based on new interval
start = newCenter - duration / 2
end = newCenter + duration / 2
if start < 0:
self.start = 0
self.end = duration
elif end > self.TextGrid.maxTime:
self.start = self.TextGrid.maxTime - duration
self.end = self.TextGrid.maxTime
else:
self.start = newCenter - duration / 2
self.end = newCenter + duration / 2
relDuration = self.end - self.start
# select new item
rel_time = newCenter - self.start
x_loc = float(rel_time / relDuration * self.canvas_width)
item = self.my_find_closest(widg, x_loc)
self.selectedItem = (widg, item)
self.setSelectedIntvlFrames(self.selectedItem)
self.fillCanvases()
self.genFrameList(widg=widg, x_loc=x_loc)
def changeTiers(self, event):
'''
'''
index = None
if self.selectedItem:
for i, el in enumerate(self.TkWidgets):
if self.selectedItem[0] in el.values():
index = i
if index != None:
if event.keysym == 'Up' and 'canvas' in self.TkWidgets[index -
1]:
new_widg = self.TkWidgets[index - 1]['canvas']
elif event.keysym == 'Down' and 'canvas' in self.TkWidgets[
index + 1]:
new_widg = self.TkWidgets[index + 1]['canvas']
else:
return
new_item = new_widg.find_withtag("frame" +
str(self.app.frame))[0]
self.selectedItem = (new_widg, new_item)
self.fillCanvases()
self.update()
self.app.Spectrogram.update()
def getMinMaxTime(self):
'''
Returns minTime and maxTime tags from selected interval
If no minTime or maxTime, returns start or end time of viewed section of TextGrid
'''
start = None
end = None
for tag in self.selectedItem[0].gettags(self.selectedItem[1]):
if tag[:7] == 'minTime':
start = float(tag[7:])
elif tag[:7] == 'maxTime':
end = float(tag[7:])
if start == None:
start = self.start
if end == None:
end = self.end
return (start, end)
def getBounds(self, event):
'''
'''
# debug(event.char, event.keysym, event.keycode)
# debug(self.app.frame)
f = self.tg_zoom_factor
a = self.end - self.start
z_out = (a - (a / f)) / 2
z_in = ((f * a) - a) / 2
old_start = self.start
old_end = self.end
if event.keysym == 'n':
if self.selectedItem:
self.start, self.end = self.getMinMaxTime()
# for tag in self.selectedItem[0].gettags(self.selectedItem[1]):
# if tag[:7] == 'minTime':
# self.start = float(tag[7:])
# elif tag[:7] == 'maxTime':
# self.end = float(tag[7:])
if event.keysym == 'a':
self.start = 0
self.end = self.TextGrid.maxTime
if event.keysym == 'o':
self.start = self.start - z_in
self.end = self.end + z_in
if event.keysym == 'i':
self.start = self.start + z_out
self.end = self.end - z_out
if event.keysym == 'Left':
start = self.start - a / (10 * f)
end = self.end - a / (10 * f)
if (start < 0):
self.start = 0
self.end = a
else:
self.start = start
self.end = end
if event.keysym == 'Right':
start = self.start + a / (10 * f)
end = self.end + a / (10 * f)
if end > self.TextGrid.maxTime:
self.start = self.TextGrid.maxTime - a
self.end = self.TextGrid.maxTime
else:
self.start = start
self.end = end
self.fillCanvases()
def getTracedFrames(self, frames):
'''
'''
frames = [frame[5:]
for frame in frames] #to get rid of word "frame in tag"
tracedFrames = []
for trace in self.app.Data.data['traces']:
tracedFrames = tracedFrames + self.app.Data.tracesExist(trace)
return set(frames).intersection(tracedFrames)
def fillCanvases(self):
'''
'''
if self.start < 0:
self.start = 0.
if self.end > self.TextGrid.maxTime:
self.end = self.TextGrid.maxTime
self.updateTimeLabels()
if self.selectedItem:
old_selected_tags = self.selectedItem[0].gettags(
self.selectedItem[1])
duration = self.end - self.start
self.frameTier = self.TextGrid.getFirst(self.frameTierName)
for el in self.TkWidgets:
if 'name' in el:
tier = self.TextGrid.getFirst(el['name'])
# debug(tier)
if 'canvas' in el:
canvas = el['canvas']
#remove previous intervals
canvas.delete('all')
#get starting interval
i = tier.indexContaining(self.start)
# not sure why, but this is sometimes None -JNW 2020-01-28
if i != None:
#debug(self.TextGrid, self.current, el, tier, self.start, i)
time = tier[i].maxTime
frame_i = 0
while i < len(tier) and tier[i].minTime <= self.end:
if self.start >= tier[i].minTime:
strtime = self.start
else:
strtime = tier[i].minTime
if self.end <= tier[i].maxTime:
time = self.end
length = time - strtime
pixel_length = length / duration * self.canvas_width
mod = length / 2
rel_time = time - self.start
loc = (rel_time - mod) / duration * self.canvas_width
text = canvas.create_text(loc,
self.canvas_height / 2,
justify='center',
text=tier[i].mark,
width=pixel_length,
activefill='blue')
minTimetag = "minTime" + str(tier[i].minTime)
maxTimetag = "maxTime" + str(tier[i].maxTime)
canvas.addtag_withtag(minTimetag, text)
canvas.addtag_withtag(maxTimetag, text)
#add containted frames to tags
while frame_i < len(
self.frameTier
) and self.frameTier[frame_i].time <= tier[i].maxTime:
if self.frameTier[frame_i].time >= tier[i].minTime:
canvas.addtag_withtag(
"frame" + self.frameTier[frame_i].mark,
text)
if tier[i].mark != '':
el['canvas-label'].addtag_all(
"frame" + self.frameTier[frame_i].mark)
frame_i += 1
#pass on selected-ness
if self.selectedItem:
if self.selectedItem[
0] != self.app.Spectrogram.canvas:
# old_selected_tags = self.selectedItem[0].gettags(self.selectedItem[1])
if minTimetag in old_selected_tags and maxTimetag in old_selected_tags and canvas == self.selectedItem[
0]:
# I'm not sure why, but when collapsing canvases this sometimes
# changes which tier is selected. Adding canvas == self.selectedItem[0]
# seems to fix this though.
# - D.S. 2020-01-30
self.selectedItem = (canvas, text)
#create line
loc = rel_time / duration * self.canvas_width
i += 1
if i < len(tier) and loc < self.canvas_width:
canvas.create_line(loc,
0,
loc,
self.canvas_height,
tags='line')
time = tier[
i].maxTime #here so that loop doesn't run an extra time
#fills labels with info about tiers w/traces
self.updateTierLabels()
elif 'frames' in el:
frames = el['frames']
i = 0
frames.delete('all')
first_frame_found = False
while i < len(tier) and tier[i].time <= self.end:
# debug(tier[i].time, i,'frame time and frame number (line 1076)')
if tier[i].time >= self.start:
# x_coord = (tier[i].time-self.start)/duration*self.canvas_width
x_coord = ((tier[i].time - self.start) *
self.canvas_width) / duration
#determine fill
if tier[i].mark in self.app.Data.getCurrentTraceTracedFrames(
):
fill = 'black'
else:
fill = 'gray70'
frame = frames.create_line(x_coord,
0,
x_coord,
self.canvas_height,
tags="frame" + tier[i].mark,
fill=fill)
if first_frame_found == False and i + 1 < len(tier):
self.firstFrame = int(tier[i].mark) + 1
first_frame_found = True
self.frame_len = tier[i + 1].time - tier[i].time
CanvasTooltip(frames, frame, text=tier[i].mark)
i += 1
self.lastFrame = int(tier[i - 1].mark)
self.paintCanvases()
if hasattr(self.app, 'Spectrogram'):
# don't try to call this during startup
# because TextGrid is loaded earlier
self.app.Spectrogram.reset()
def updateTimeLabels(self):
'''
'''
self.current = self.TextGrid.getFirst(
self.frameTierName)[self.app.frame - 1].time
self.TkWidgets[-1]['times'].itemconfig(1,
text='{:.6f}'.format(
self.start))
self.TkWidgets[-1]['times'].itemconfig(2,
text='{:.6f}'.format(self.end))
self.TkWidgets[-1]['times'].itemconfig(3,
text='{:.6f}'.format(
self.current))
def updateTierLabels(self):
'''
'''
for el in self.TkWidgets:
if 'canvas' in el:
current_label = el['canvas-label'].find_all()[0]
nonempty_frames = el['canvas-label'].gettags(current_label)
el['canvas-label'].itemconfig(
current_label,
text='{}:\n({}/{})'.format(
el['name'], len(self.getTracedFrames(nonempty_frames)),
len(nonempty_frames)))
def my_find_closest(self, widg, x_loc):
'''
replaces TkInter's find_closest function, which is buggy, determines
whether found item is text, line, or label, and returns corresponding item
'''
#could be more efficient FIXME
maybe_item = None
dist = 999999999999
for el in widg.find_all():
obj_x = widg.coords(el)[0]
if abs(obj_x - x_loc) < dist:
dist = abs(obj_x - x_loc)
maybe_item = el
if widg in self.tier_pairs.keys(): #on tier-label canvas
#fill selected tier frames
# self.setSelectedIntvlFrames(widg,item)
item = maybe_item
elif widg in self.tier_pairs.values(
): #on canvas with intervals/frames
if isinstance(maybe_item, int):
# #if item found is a boundary
# if len(widg.gettags(maybe_item)) == 0 or widg.gettags(maybe_item) == ('current',):
if 'line' in widg.gettags(maybe_item):
#determine on which side of the line the event occurred
if widg.coords(maybe_item)[0] > x_loc:
item = maybe_item - 1
else: #i.e. event was on line or to the right of it
item = maybe_item + 1
else:
item = maybe_item
# self.setSelectedIntvlFrames(widg,item)
else:
item = maybe_item
return item
def setSelectedIntvlFrames(self, tupl):
''' '''
widg, item = tupl
self.selectedIntvlFrames = []
for x in widg.gettags(item):
if x[:5] == 'frame':
self.selectedIntvlFrames.append(x[5:])
def wipeFill(self):
'''
Turns selected frame and interval back to black
'''
for frame in range(1, self.app.frames + 1):
if str(frame) in self.app.Data.getCurrentTraceTracedFrames():
fill = 'black'
else:
fill = 'gray70'
self.frames_canvas.itemconfig('frame' + str(frame), fill=fill)
if self.selectedItem:
wdg, itm = self.selectedItem
if wdg.type(itm) != 'text' and wdg.type(itm) != None:
wdg, itm = self.app.Spectrogram.oldSelected
wdg.itemconfig(itm, fill='black')
if len(wdg.find_withtag(itm + 1)) > 0:
wdg.itemconfig(itm + 1, fill='black')
if len(wdg.find_withtag(itm - 1)) > 0:
wdg.itemconfig(itm - 1, fill='black')
#clicked tier label
if wdg in self.tier_pairs.keys():
wdg.itemconfig(1, fill='black')
self.tier_pairs[wdg].itemconfig('all', fill='black')
self.frames_canvas.itemconfig('all', fill='black')
def genFrameList(self, event=None, widg=None, x_loc=None, SI=False):
'''
Upon click, reads frames within interval from the tags to the text item of that interval,
and highlights text of clicked interval
'''
self.wipeFill()
if event:
widg = event.widget
x_loc = event.x
if SI == False:
item = self.my_find_closest(widg, x_loc)
self.selectedItem = (widg, item)
self.setSelectedIntvlFrames(self.selectedItem)
#automatically updates frame
if not str(self.app.frame) in self.selectedIntvlFrames:
if self.selectedIntvlFrames:
new_frame = int(self.selectedIntvlFrames[0])
else:
frame = self.my_find_closest(self.frames_canvas, x_loc)
framenum = self.frames_canvas.gettags(frame)[0][5:]
new_frame = int(framenum)
if self.firstFrame > new_frame:
new_frame = self.firstFrame
elif new_frame > self.lastFrame:
new_frame = self.lastFrame
self.app.frame = new_frame
self.app.framesUpdate()
else:
self.paintCanvases()
self.app.Spectrogram.update()
def collapse(self, event):
'''
collapse or uncollapse selected tier
'''
widg = event.widget
if event.num == 1:
h = self.collapse_height
if int(widg['height']) == h:
h = self.canvas_height
elif event.num == 4 or event.delta > 0:
h = self.canvas_height
else:
h = self.collapse_height
if int(widg['height']) == h:
return
elif h == self.canvas_height:
mv = (self.canvas_height - self.collapse_height - 14) / 2
else:
mv = (self.collapse_height + 14 - self.canvas_height) / 2
# manually shifting the text by 7 pixels is a rather ugly hack,
# but it works - DS
if widg in self.tier_pairs:
l, c = widg, self.tier_pairs[widg]
else:
c = widg
l = None
for k in self.tier_pairs:
if self.tier_pairs[k] == widg:
l = k
break
l.configure(height=h)
c.configure(height=h)
l.move('all', 0, mv)
self.app.event_generate('<Configure>')
def paintCanvases(self):
'''
'''
if self.selectedItem:
wdg, itm = self.selectedItem
#paint selected
if wdg.type(itm) == 'text':
wdg.itemconfig(itm, fill='blue')
#paint boundaries of selected
if itm + 1 in wdg.find_all():
wdg.itemconfig(itm + 1, fill='blue')
if itm - 1 in wdg.find_all():
wdg.itemconfig(itm - 1, fill='blue')
if wdg in self.tier_pairs.keys(): #if on tier-label canvas
canvas = self.tier_pairs[wdg]
for el in canvas.find_all():
# #make all text intervals blue
# if canvas.type(canvas.find_withtag(el)) == 'text':
canvas.itemconfig(el, fill='blue')
#paint frames
frames = wdg.gettags(itm)
for frame in frames:
if frame[:5] == 'frame':
frame_obj = self.frames_canvas.find_withtag(frame)
#detect whether frame contains any traces
framenum = frame[5:]
if framenum in self.app.Data.getCurrentTraceTracedFrames():
fill = 'blue'
else:
fill = 'dodger blue'
self.frames_canvas.itemconfig(frame_obj, fill=fill)
#current frame highlighted in red
if self.app.frame:
self.highlighted_frame = self.frames_canvas.find_withtag(
'frame' + str(self.app.frame))
self.frames_canvas.itemconfig(self.highlighted_frame, fill='red')
def update(self):
'''
'''
# debug(self.frames_canvas)
#create list of displayed frames' tags
itrobj = []
for itm in self.frames_canvas.find_all():
itrobj += list(self.frames_canvas.gettags(itm))
#if selected frame is out of view
if "frame" + str(self.app.frame) not in itrobj:
duration = self.end - self.start
#recenter view on selected frame
new_time = self.TextGrid.getFirst(
self.frameTierName)[self.app.frame - 1].time
self.start = new_time - (duration / 2)
self.end = new_time + (duration / 2)
#redraw
self.fillCanvases()
self.wipeFill()
#if selected frame outside selected interval, select interval on same tier containing frame
if self.selectedItem:
if self.selectedItem[0] in self.tier_pairs.keys(
) or self.selectedItem[0] in self.tier_pairs.values():
if "frame" + str(
self.app.frame) not in self.selectedItem[0].gettags(
self.selectedItem[1]
): #FIXME should also detect if on label canvas
widg = self.selectedItem[0]
if widg in self.tier_pairs:
widg = self.tier_pairs[widg]
new_interval = widg.find_withtag("frame" +
str(self.app.frame))[0]
self.selectedItem = (self.selectedItem[0], new_interval)
# repaint all frames
self.paintCanvases()
self.updateTimeLabels()
def grid(self, event=None):
'''
Wrapper for gridding all of our Tk widgets. This funciton assumes that the tiers (as
specified in the actual TextGrid files) are in some sort of reasonable order, with the
default label being drawn on top.
'''
for t in range(len(self.TkWidgets)):
tierWidgets = self.TkWidgets[t]
if 'label' in tierWidgets:
tierWidgets['label'].grid(row=t, column=0, sticky='w')
if 'frames' in tierWidgets:
tierWidgets['frames'].grid(row=t,
column=2,
sticky='w',
pady=self.app.pady)
tierWidgets['frames-label'].grid(row=t,
column=0,
sticky='w',
pady=self.app.pady)
if 'canvas' in tierWidgets:
tierWidgets['canvas'].grid(row=t,
column=2,
sticky='w',
pady=self.app.pady / 2)
tierWidgets['canvas-label'].grid(row=t,
column=0,
sticky='w',
pady=self.app.pady / 2)
self.tier_pairs[
tierWidgets['canvas-label']] = tierWidgets['canvas']
if 'times' in tierWidgets:
tierWidgets['times'].grid(row=t, column=2, sticky='s')
def grid_remove(self):
raise NotImplementedError('cannot grid_remove the TextGridModule')
def openSearch(self, event=None):
self.app.Search.openSearch()
def unnorm(norm_value, min, max):
return norm_value * (max - min) + min
seg_ind = 0
for f in sorted(os.listdir(data_dir)):
if not f.endswith('.TextGrid'):
continue
print(f)
wav_file = f.replace('.TextGrid', '.adc.wav')
textgrid_path = os.path.join(data_dir, f)
wav_path = os.path.join(data_dir, wav_file)
tg = TextGrid()
tg.read(textgrid_path)
word_tier = tg.getFirst('words')
segmentation_tier = IntervalTier('segments', 0, word_tier.maxTime)
durations = []
for interval in word_tier:
if interval.mark == '':
continue
durations.append(interval.maxTime - interval.minTime)
max_duration = max(durations)
min_duration = min(durations)
min_thresh = 0.01
max_thresh = 0.05
segs = []
for interval in word_tier:
def unnorm(norm_value, min, max):
return norm_value * (max - min) + min
seg_ind = 0
for f in sorted(os.listdir(data_dir)):
if not f.endswith('.TextGrid'):
continue
print(f)
wav_file = f.replace('.TextGrid', '.adc.wav')
textgrid_path = os.path.join(data_dir, f)
wav_path = os.path.join(data_dir, wav_file)
tg = TextGrid()
tg.read(textgrid_path)
word_tier = tg.getFirst('words')
segmentation_tier = IntervalTier('segments', 0, word_tier.maxTime)
durations = []
for interval in word_tier:
if interval.mark == '':
continue
durations.append(interval.maxTime - interval.minTime)
max_duration = max(durations)
min_duration = min(durations)
min_thresh = 0.01
max_thresh = 0.05
segs = []
for interval in word_tier:
r_code, bnc_code = tg_path.split('_')[-3:-1]
if bnc_code not in bnc_cache:
bnc_cache[bnc_code] = load_bnc_code(bnc_code)
speakers.update(bnc_cache[bnc_code][0])
_, recording_data, transcripts = bnc_cache[bnc_code]
transcript = transcripts[r_code]
tg = TextGrid(strict=False)
try:
tg.read(tg_path)
except:
print('Error reading {}'.format(tg_path))
#print(tg.minTime, tg.maxTime, tg.maxTime - tg.minTime)
analysis.append(
[f, duration,
os.path.basename(tg_path), tg.minTime, tg.maxTime])
word_tier = tg.getFirst('word')
#print([x.mark for x in word_tier])
phone_tier = tg.getFirst('phone')
print(
'There were {} read errors, {} duplicated mins, {} duplicated maxs, and {} overlaps.'
.format(read_errors, dup_min_errors, dup_max_errors, overlap_errors))
error
with open(os.path.join(base_dir, 'analysis.txt'), 'w') as f:
writer = csv.writer(f)
writer.writerow(['wav', 'duration', 'tg', 'tg_min', 'tg_max'])
for line in analysis:
writer.writerow(line)
error
