def __init__(self,beta=10.0):
self.CMU,self.CMUX = load_CMU()
self.A = ARPAstat(1.0)
#self.min_cutoff = 1.25
self.min_cutoff = 1.50
self.beta = beta
class homophonic_word_translate(object):
def __init__(self,beta=10.0):
self.CMU,self.CMUX = load_CMU()
self.A = ARPAstat(1.0)
#self.min_cutoff = 1.25
self.min_cutoff = 1.50
self.beta = beta
def __call__(self,w1):
if w1 not in self.CMU:
return w1
c1 = self.CMU[w1]
CLOSE = {}
for w2 in self.CMUX[len(c1)]:
if w2 not in VALID_WORDS: continue
c2 = self.CMU[w2]
dx = sum([self.A.delta(x,y) for x,y in zip(c1,c2)]) / len(c1)
if dx < self.min_cutoff and w1!=w2:
CLOSE[w2] = dx
# No match was found, return original wor
if not CLOSE:
return w1
WORDS = CLOSE.keys()
E = np.array([CLOSE[word] for word in WORDS])
Z = np.exp(-E*self.beta)
prob = np.exp(-E*self.beta)/Z.sum()
return np.random.choice(WORDS, p=prob)
