Produce candidate pronunciation for an acronym. Take set of letters, perhaps a potential acronym, and try to expand it. Maybe just to have a new word to pronounce. Turn crps to carapace.

acronym2word.py

import numpy as np
import toolz as tz

np.random.seed(1000)

# String -> String 
def generate_word(word):
    """ Given a word, return a new word with vowels interspersed between the letters. """
    vowels = list('aeiouy_')
    letters = list(word)
    additions = list(np.random.choice(vowels, size = len(letters) + 1, replace = True))
    return (''.join(list(tz.interleave([additions, letters])))).replace('_','')

[generate_word('crps') for _ in range(20)]

Doesn't work very well. Would be better to actually break into phonemes and construct words from common phonemes, or to find words that seem to contain these phonemes. Could be made to work, I think. Perhaps NLTK has some tools that could assist in this.