def validate(self):
"""Ensures that adjacent start/stop times are valid by adding silence"""
i = 0
start = 0.
while i < len(self) - 1:
# Get start and end times between words
end = self[i].start()
# Patch gap with silence
if end - start > 1e-3:
# Extend existing silence if possible
if str(self[i]) == pypar.SILENCE:
self[i][0]._start = start
else:
word = pypar.Word(
pypar.SILENCE,
[pypar.Phoneme(pypar.SILENCE, start, end)])
self._words.insert(i, word)
i += 1
i += 1
start = self[i].end()
# Phoneme gap validation
for word in self:
word.validate()
def load_textgrid(self, filename):
"""Load from textgrid file"""
# Load file
grid = textgrid.TextGrid.fromFile(filename)
# Get phoneme and word representations
phon_tier, word_tier = grid[0], grid[1]
# Iterate over words
words = []
phon_idx = 0
for word in word_tier:
# Get all phonemes for this word
phonemes = []
while phon_idx < len(phon_tier) and \
phon_tier[phon_idx].maxTime <= word.maxTime:
phoneme = phon_tier[phon_idx]
mark = phoneme.mark if phoneme.mark != 'sil' else pypar.SILENCE
phonemes.append(
pypar.Phoneme(mark, phoneme.minTime, phoneme.maxTime))
phon_idx += 1
# Add finished word
words.append(pypar.Word(word.mark, phonemes))
return words
def parse_json(self, alignment):
"""Construct word list from json representation"""
words = []
for word in alignment['words']:
try:
# Add a word
phonemes = [
pypar.Phoneme(*phoneme) for phoneme in word['phonemes']
]
words.append(pypar.Word(word['alignedWord'], phonemes))
except KeyError:
# Add a silence
phonemes = [
pypar.Phoneme(pypar.SILENCE, word['start'], word['end'])
]
words.append(pypar.Word(pypar.SILENCE, phonemes))
return words
def validate(self):
"""Ensures that adjacent start/end times are valid by adding silence"""
i = 0
while i < len(self) - 1:
# Get start and end times between phonemes
start = self[i].end()
end = self[i + 1].start()
# Patch gap with silence
if end - start > 1e-4:
phoneme = pypar.Phoneme(pypar.SILENCE, start, end)
self.phonemes.insert(i + 1, phoneme)
i += 1
i += 1
def load_mlf(self, filename):
"""Load from mlf file"""
# Load file from disk
with open(filename) as file:
# Read in phoneme alignment
lines = [
Line(line) for line in file.readlines()
if self.line_is_valid(line)
]
# Remove silence tokens with 0 duration
lines = [line for line in lines if line.start < line.end]
# Extract words and phonemes
phonemes = []
words = []
for line in lines:
# Start new word
if line.word is not None:
# Add word that just finished
if phonemes:
words.append(pypar.Word(word, phonemes))
phonemes = []
word = line.word
# Add a phoneme
phonemes.append(pypar.Phoneme(line.phoneme, line.start, line.end))
# Handle last word
if phonemes:
words.append(pypar.Word(word, phonemes))
return words