def generate_phonemes(self):
"""
Step 3. Generate phoneme list from word list. Map phonemes to single characters to
simplify encoding issues.
"""
self.log.debug("G2P.generate_phonemes()")
phones = []
for w in self.dic:
for p in w.phonemes:
if p not in phones:
phones.append(p)
(self.pmap_c2s, self.pmap_s2c) = charmap.create_char_maps(phones)
for w in self.dic:
for i in range(len(w.phonemes)):
w.phonemes[i] = self.pmap_s2c[w.phonemes[i]]
self.phonemes = self.pmap_c2s.keys()
def generate_graphemes(self):
"""
Step 2. Generate grapheme list from word list. Map graphemes to single characters to
simplify encoding issues.
"""
self.log.debug("G2P.generate_graphemes()")
graphs = []
for w in self.dic:
for g in w.graphemes:
if g not in graphs:
graphs.append(g)
(self.gmap_c2s, self.gmap_s2c) = charmap.create_char_maps(graphs)
for w in self.dic:
for i in range(len(w.graphemes)):
w.graphemes[i] = self.gmap_s2c[w.graphemes[i]]
# Add null grapheme possibility
self.graphemes = [ '0' ]
self.graphemes.extend( self.gmap_c2s.keys())
