ConstantineLignos

from collections import Counter
import nltk

TOP_PERCENT = .01

def prob_mass_top(counts, n):
    return sum(count for word, count in counts.most_common(n)) / float(sum(count.values()))

count = Counter(word.lower() for word in nltk.corpus.brown.words())

ConstantineLignos

import re
from collections import defaultdict

import nltk
from nltk.corpus import cmudict

def clean_pron(pron):
    """Remove stress from pronunciations."""
    return re.sub(r"\d", "", pron)