def create_text_grid(text_grid_path, label, num_of_frames):
"""
create TextGrid files from the labels_path file
:param num_of_frames:
:param label:
:param text_grid_path:
"""
# should be different for every file format
FRAME_RATE = 10
MILLISEC_2_SEC = 0.001
FRAME_CONCAT_ONSET = 0
FRAME_CONCAT_OFFSET = 0
v_onset = (label[0] + FRAME_CONCAT_ONSET)
v_offset = (label[1] + FRAME_CONCAT_OFFSET)
length = num_of_frames * FRAME_RATE * MILLISEC_2_SEC
# build TextGrid
text_grid = TextGrid()
vowels_tier = IntervalTier(name='Duration', minTime=0.0, maxTime=float(length))
vowels_tier.addInterval(Interval(0, float(v_onset) * FRAME_RATE * MILLISEC_2_SEC, ""))
vowels_tier.addInterval(
Interval(float(v_onset) * FRAME_RATE * MILLISEC_2_SEC, float(v_offset) * FRAME_RATE * MILLISEC_2_SEC, ""))
vowels_tier.addInterval(Interval(float(v_offset) * FRAME_RATE * MILLISEC_2_SEC, float(length), ""))
text_grid.append(vowels_tier)
text_grid.write(text_grid_path)
def create_text_grid(text_grid_path, label, num_of_frames):
"""
create TextGrid files from the labels_path file
:param num_of_frames:
:param label:
:param text_grid_path:
"""
# should be different for every file format
FRAME_RATE = 10
MILLISEC_2_SEC = 0.001
FRAME_CONCAT_ONSET = 0
FRAME_CONCAT_OFFSET = 0
v_onset = (label[0] + FRAME_CONCAT_ONSET)
v_offset = (label[1] + FRAME_CONCAT_OFFSET)
length = num_of_frames * FRAME_RATE * MILLISEC_2_SEC
# build TextGrid
text_grid = TextGrid()
vowels_tier = IntervalTier(name='Duration',
minTime=0.0,
maxTime=float(length))
vowels_tier.addInterval(
Interval(0,
float(v_onset) * FRAME_RATE * MILLISEC_2_SEC, ""))
vowels_tier.addInterval(
Interval(
float(v_onset) * FRAME_RATE * MILLISEC_2_SEC,
float(v_offset) * FRAME_RATE * MILLISEC_2_SEC, ""))
vowels_tier.addInterval(
Interval(
float(v_offset) * FRAME_RATE * MILLISEC_2_SEC, float(length), ""))
text_grid.append(vowels_tier)
text_grid.write(text_grid_path)
def create_db(audio_path, textgrid_path, output_path):
voiced_path = output_path + 'voiced/'
prevoiced_path = output_path + 'prevoiced/'
c = 0
for item in os.listdir(audio_path):
tg_file_path = textgrid_path + item.replace('.wav', '.TextGrid')
if item.endswith('.wav') and os.path.exists(tg_file_path):
c += 1
print('Processing item: %d, file name: %s' % (c, item))
tg = TextGrid()
tg.read(tg_file_path)
tier = tg._TextGrid__tiers[1]
print('Creating output dirs ...')
# create the relevant dirs
if not os.path.exists(voiced_path):
os.mkdir(voiced_path)
if not os.path.exists(prevoiced_path):
os.mkdir(prevoiced_path)
for i, interval in enumerate(tier._IntervalTier__intervals):
if 'ne' in interval._Interval__mark:
# gap = 0.02
start_vot = interval._Interval__xmin
end_vot = interval._Interval__xmax
gap = round((end_vot - start_vot) * 2, 2)
gap = max(0.05, gap)
start = max(0, start_vot - gap)
end = end_vot + gap
output_name = os.path.abspath(prevoiced_path) + '/' + str(
i) + '_' + item
utils.crop_wav(os.path.abspath(audio_path + item), start,
end, output_name)
# write the text grid
length = end - start
new_tg = TextGrid()
vot_tier = IntervalTier(name='VOT',
xmin=0.0,
xmax=float(length))
vot_tier.append(Interval(0, start_vot - start, ""))
vot_tier.append(
Interval(start_vot - start, end_vot - start, "neg"))
vot_tier.append(
Interval(end_vot - start, float(length), ""))
new_tg.append(vot_tier)
new_tg.write(output_name.replace('.wav', '.TextGrid'))
elif 'v' in interval._Interval__mark:
# gap = 0.02
start_vot = interval._Interval__xmin
end_vot = interval._Interval__xmax
gap = round((end_vot - start_vot) * 2, 2)
gap = max(0.05, gap)
start = max(0, start_vot - gap)
end = end_vot + gap
output_name = os.path.abspath(voiced_path) + '/' + str(
i) + '_' + item
utils.crop_wav(os.path.abspath(audio_path + item), start,
end, output_name)
# write the text grid
length = end - start
new_tg = TextGrid()
vot_tier = IntervalTier(name='VOT',
xmin=0.0,
xmax=float(length))
vot_tier.append(Interval(0, start_vot - start, ""))
vot_tier.append(
Interval(start_vot - start, end_vot - start, "v"))
vot_tier.append(
Interval(end_vot - start, float(length), ""))
new_tg.append(vot_tier)
new_tg.write(output_name.replace('.wav', '.TextGrid'))
print('Done.')
return None
end = t._TextGrid__xmax
length = end - start
# ========= merge and save ========= #
text_grid = TextGrid()
# == DCM == #
onset = orig_manual_annotations[i][1] - (manual_annotations[i][1] -
dcm_annotations[i][1])
offset = orig_manual_annotations[i][2] - (manual_annotations[i][2] -
dcm_annotations[i][2]) - 0.01
dcm_tier = IntervalTier(name='DCM', xmin=0.0, xmax=float(length))
dcm_tier.append(Interval(0, float(onset), ""))
dcm_tier.append(Interval(float(onset), float(offset), "vowel"))
dcm_tier.append(Interval(float(offset), float(length), ""))
text_grid.append(dcm_tier)
# == DCM NC == #
onset = orig_manual_annotations[i][1] - (manual_annotations[i][1] -
dcm_nc_annotations[i][1])
offset = orig_manual_annotations[i][2] - (
manual_annotations[i][2] - dcm_nc_annotations[i][2]) - 0.01
dcm_nc_tier = IntervalTier(name='DCM-NC', xmin=0.0, xmax=float(length))
dcm_nc_tier.append(Interval(0, float(onset), ""))
dcm_nc_tier.append(Interval(float(onset), float(offset), "vowel"))
dcm_nc_tier.append(Interval(float(offset), float(length), ""))
text_grid.append(dcm_nc_tier)
# == FAVE == #
onset = orig_manual_annotations[i][1] - (manual_annotations[i][1] -
fave_annotations[i][1])
t.read(manual_path_orig+orig_manual_annotations[i][0])
start = t._TextGrid__xmin
end = t._TextGrid__xmax
length = end - start
# ========= merge and save ========= #
text_grid = TextGrid()
# == DCM == #
onset = orig_manual_annotations[i][1] - (manual_annotations[i][1] - dcm_annotations[i][1])
offset = orig_manual_annotations[i][2] - (manual_annotations[i][2] - dcm_annotations[i][2]) - 0.01
dcm_tier = IntervalTier(name='DCM', xmin=0.0, xmax=float(length))
dcm_tier.append(Interval(0, float(onset), ""))
dcm_tier.append(Interval(float(onset), float(offset), "vowel"))
dcm_tier.append(Interval(float(offset), float(length), ""))
text_grid.append(dcm_tier)
# == DCM NC == #
onset = orig_manual_annotations[i][1] - (manual_annotations[i][1] - dcm_nc_annotations[i][1])
offset = orig_manual_annotations[i][2] - (manual_annotations[i][2] - dcm_nc_annotations[i][2]) - 0.01
dcm_nc_tier = IntervalTier(name='DCM-NC', xmin=0.0, xmax=float(length))
dcm_nc_tier.append(Interval(0, float(onset), ""))
dcm_nc_tier.append(Interval(float(onset), float(offset), "vowel"))
dcm_nc_tier.append(Interval(float(offset), float(length), ""))
text_grid.append(dcm_nc_tier)
# == FAVE == #
onset = orig_manual_annotations[i][1] - (manual_annotations[i][1] - fave_annotations[i][1])
offset = orig_manual_annotations[i][2] - (manual_annotations[i][2] - fave_annotations[i][2]) - 0.01
fave_tier = IntervalTier(name='FAVE', xmin=0.0, xmax=float(length))
fave_tier.append(Interval(0, float(onset), ""))
def create_db(audio_path, textgrid_path, output_path):
voiced_path = output_path+'voiced/'
prevoiced_path = output_path+'prevoiced/'
c = 0
for item in os.listdir(audio_path):
tg_file_path = textgrid_path + item.replace('.wav', '.TextGrid')
if item.endswith('.wav') and os.path.exists(tg_file_path):
c += 1
print('Processing item: %d, file name: %s' % (c, item))
tg = TextGrid()
tg.read(tg_file_path)
tier = tg._TextGrid__tiers[1]
print('Creating output dirs ...')
# create the relevant dirs
if not os.path.exists(voiced_path):
os.mkdir(voiced_path)
if not os.path.exists(prevoiced_path):
os.mkdir(prevoiced_path)
for i, interval in enumerate(tier._IntervalTier__intervals):
if 'ne' in interval._Interval__mark:
# gap = 0.02
start_vot = interval._Interval__xmin
end_vot = interval._Interval__xmax
gap = round((end_vot - start_vot)*2, 2)
gap = max(0.05, gap)
start = max(0, start_vot - gap)
end = end_vot + gap
output_name = os.path.abspath(prevoiced_path) + '/' + str(i) + '_' + item
utils.crop_wav(os.path.abspath(audio_path + item), start, end,
output_name)
# write the text grid
length = end - start
new_tg = TextGrid()
vot_tier = IntervalTier(name='VOT', xmin=0.0, xmax=float(length))
vot_tier.append(Interval(0, start_vot - start, ""))
vot_tier.append(Interval(start_vot - start, end_vot - start, "neg"))
vot_tier.append(Interval(end_vot - start, float(length), ""))
new_tg.append(vot_tier)
new_tg.write(output_name.replace('.wav', '.TextGrid'))
elif 'v' in interval._Interval__mark:
# gap = 0.02
start_vot = interval._Interval__xmin
end_vot = interval._Interval__xmax
gap = round((end_vot - start_vot)*2, 2)
gap = max(0.05, gap)
start = max(0, start_vot - gap)
end = end_vot + gap
output_name = os.path.abspath(voiced_path) + '/' + str(i) + '_' + item
utils.crop_wav(os.path.abspath(audio_path + item), start, end,
output_name)
# write the text grid
length = end - start
new_tg = TextGrid()
vot_tier = IntervalTier(name='VOT', xmin=0.0, xmax=float(length))
vot_tier.append(Interval(0, start_vot - start, ""))
vot_tier.append(Interval(start_vot - start, end_vot - start, "v"))
vot_tier.append(Interval(end_vot - start, float(length), ""))
new_tg.append(vot_tier)
new_tg.write(output_name.replace('.wav', '.TextGrid'))
print('Done.')
return None
