CMU Dictionary and Rhyme
04 Apr 2020
This is just a casual Python project to explore using the CMU Dictionary to find rhymes and other phonetic similarities between words. I mostly built off of the the pronouncing library.
For now I’ll just paste the code below and link Github .
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
import random
import random
import string
import pronouncing
import re
import itertools
import numpy as np
"""
Web links for more information:
http://www.speech.cs.cmu.edu/cgi-bin/cmudict
https://en.wikipedia.org/wiki/Arpabet
https://en.wikipedia.org/wiki/Rhyme
https://en.wikipedia.org/wiki/Perfect_and_imperfect_rhymes
https://en.wikipedia.org/wiki/Assonance
https://www.litcharts.com/literary-devices-and-terms/slant-rhyme
"""
def consonant_clusters():
""" Return a list of possible English consonant clusters."""
"""
See the following resources for more information on consonant clusters.
"consonant cluster: "a group of consonants which have no intervening vowel"
https://en.wikipedia.org/wiki/Consonant_cluster
"Two Theories of Onset Clusters", Duanmu
http://www-personal.umich.edu/~duanmu/CR02.pdf
"Phoneme distribution and syllable structure of entry words in the CMU
English Pronouncing Dictionary", Yang
http://fonetiks.info/bgyang/db/201606cmu.pdf
"Blends, Digraphs, Trigraphs, and Other Letter Combinations"
https://www.enchantedlearning.com/consonantblends/
"""
return ['F W', 'F R', 'F L', 'S W', 'S V',
'S R', 'S L', 'S N', 'S M', 'S F',
'S P', 'S T', 'S K', 'SH W', 'SH R',
'SH L', 'SH N', 'SH M', 'TH W', 'TH R',
'V W', 'V R', 'V L', 'Z W', 'Z L',
'B W', 'B R', 'B L', 'D W', 'D R',
'G W', 'G R', 'G L', 'P W', 'P R',
'P L', 'T W', 'T R', 'K W', 'K R',
'K L', 'L Y', 'N Y', 'M Y', 'V Y',
'H Y', 'F Y', 'S Y', 'TH Y', 'Z Y',
'B Y', 'D Y', 'G Y', 'P Y', 'T Y',
'K Y', 'S P L', 'S P R', 'S T R', 'S K R',
'S K W']
def check_if_consonant_cluster(phones):
"""Return True if CMUdict phonemes is a consonant cluster."""
return phones in consonant_clusters()
def check_if_vowel(phone):
"""Returns True if CMUdict phoneme is a vowel."""
# all vowels in CMU Pronouncing Dictionary have stress number 0-2
return phone[-1] in '012'
def check_if_stressed_vowel(phone):
"""Returns True if CMUdict phoneme is a stressed vowel."""
# 1 or 2 indicate vowel is stressed
return phone[-1] in '12'
def check_if_non_stressed_vowel(phone):
"""Returns True if CMUdict phoneme is a non-stressed vowel."""
# 0 indicates vowel is unstressed
return phone[-1] in '0'
def check_if_consonant(phone):
"""Returns True if CMUdict phoneme is a consonant."""
# consonants do not have any stress number
return phone[-1] not in '012'
def unique(data_list):
"""Removes duplicates from a list, i.e. just unique elements."""
return list(dict.fromkeys(data_list))
def all_the_same(data_list, val):
"""Returns true if all elements in data_list at equal to val"""
if len(data_list) > 0 and len(data_list) == data_list.count(val):
return True
else:
return False
def random_phones_for_word(word):
"""Chooses random set of CMUdict phonemes for word
:param word: a word
:return: CMUdict phonemes string
"""
all_phones = pronouncing.phones_for_word(word)
if not all_phones:
return ""
phones = random.choice(all_phones)
return phones
def first_phones_for_word(word):
"""Chooses first set of CMUdict phonemes for word
:param word: a word
:return: CMUdict phonemes string
"""
all_phones = pronouncing.phones_for_word(word)
if not all_phones:
return ""
phones = all_phones[0]
return phones
def rhyme(word, phones=None):
""" Returns a list of rhymes for a word.
The conditions for this 'normal' rhyme between words are:
(1) last stressed vowel and subsequent phonemes match
If phones argument not given, phones/pronunciation used will default to the
first in the list of phones returned for word. If no rhyme is found, an
empty list is returned.
This is the 'default' rhyme, same definition used by the pronoucning
module for its 'rhymes' function. This is also like the shared set of
perfect and identical rhymes, except the identical word will be removed
from the returned rhymes list.
:param word: a word
:param phones: specific CMUdict phonemes string for word (default None)
:return: a rhyme for word
"""
if phones is None:
phones = first_phones_for_word(word)
if phones == "":
return []
else:
if phones not in pronouncing.phones_for_word(word):
raise ValueError(phones + " not phones for " + word)
if not phones:
raise ValueError("phonemes string is empty")
return [
w for w in
pronouncing.rhyme_lookup.get(pronouncing.rhyming_part(phones), [])
if (w != word)]
def perfect_rhyme(word, phones=None):
""" Returns a list of perfect rhymes for a word.
The conditions for a perfect rhyme between words are:
(1) last stressed vowel and subsequent phonemes match
(2) onset of last stressed syllable is different
If phones argument not given, phones/pronunciation used will default to the
first in the list of phones returned for word. If no rhyme is found, an
empty list is returned.
:param word: a word
:param phones: specific CMUdict phonemes string for word (default None)
:return: a list of perfect rhymes for word
"""
if phones is None:
phones = first_phones_for_word(word)
if phones == "":
return []
else:
if phones not in pronouncing.phones_for_word(word):
raise ValueError(phones + " not phones for +" + word)
if not phones:
raise ValueError("phonemes string is empty")
perf_and_iden_rhymes = rhyme(word, phones)
identical_rhymes = identical_rhyme(word, phones)
perfect_rhymes = list(np.setdiff1d(perf_and_iden_rhymes, identical_rhymes))
if word in perfect_rhymes:
perfect_rhymes.remove(word)
return perfect_rhymes
def identical_rhyme(word, phones=None):
""" Returns identical rhymes of word.
The conditions for an identical rhyme between words are:
(1) last stressed vowel and subsequent phonemes match
(2) onset of last stressed syllable is the same
e.g. 'leave' and 'leave', or 'leave' and 'believe'
If phones argument not given, phones/pronunciation used will default to the
first in the list of phones returned for word. If no rhyme is found, an
empty list is returned.
The identical part of the word doesn't have to be a 'real' word.
e.g. The phonemes for 'vection' will be used to find identical rhymes
of 'convection' (e.g. advection) even though 'vection' is unusual/obscure.
:param word: a word
:param phones: specific CMUdict phonemes string for word (default None)
:return: a list of identical rhymes for word
"""
if phones is None:
phones = first_phones_for_word(word)
if phones == "":
return []
else:
if phones not in pronouncing.phones_for_word(word):
raise ValueError(phones + " not phones for +" + word)
if not phones:
raise ValueError("phonemes string is empty")
phones_list = phones.split()
search_list = []
for i in range(len(phones_list)-1, -1, -1):
phone = phones_list[i]
if check_if_stressed_vowel(phone) is False:
search_list.append(phone)
else:
search_list.append(phone)
last_stressed_vowel_at_start = (i == 0)
if last_stressed_vowel_at_start is True:
search_list.reverse()
search = ' '.join(search_list)
rhymes = pronouncing.search(search + "$")
return rhymes
else:
consonant_cnt = 0
consonants= ""
search_start = ""
for j in range(i, 0, -1):
next_phone = phones_list[j-1]
if check_if_consonant(next_phone) is True:
consonant_cnt += 1
if consonant_cnt > 1:
consonants = next_phone + " " + consonants
if check_if_consonant_cluster(consonants):
search_list.append(next_phone)
else:
break
else:
consonants = next_phone
search_list.append(next_phone)
else:
if consonant_cnt == 0: # null onset
# Regex: vowel (AA1, EH0, ect.) or start '^'
# pretty sure all vowel start two letters...
# (would be "((.{1,2}(0|1|2) )|^)" otherwise)
search_start = "((..(0|1|2) )|^)"
break
search_list.reverse()
search = search_start + ' '.join(search_list) + "$"
rhymes = pronouncing.search(search)
rhymes = unique(rhymes)
# for r in rhymes:
# print(pronouncing.phones_for_word(r)[0])
return rhymes
def near_rhyme(word, phones=None, stress=True, consonant_tail=0):
""" Returns a list of words that almost rhyme
The conditions for a near rhyme between words are:
(1) At least one of the phonemes after and including the last stressed
syllable match, except for the case where they all do.
If phones argument not given, phones/pronunciation used will default to the
first in the list of phones returned for word. If no rhyme is found, an
empty list is returned.
:param word: a word
:param phones: specific CMUdict phonemes string for word (default None)
:param stress: if vowels will match stress (default True)
:param consannt_tail: number of
:return: a list of near rhymes for word
"""
if phones is None:
phones = first_phones_for_word(word)
if phones == "":
return []
else:
if phones not in pronouncing.phones_for_word(word):
raise ValueError(phones + " not phones for" + word)
if not phones:
raise ValueError("phonemes string is empty")
rp = pronouncing.rhyming_part(phones)
search_combos = wildcard_mix_phones_regex_searches(rp, stress)
rhymes = []
for search in search_combos:
rhymes += pronouncing.search(
search + "( .{1,3}){0," + str(consonant_tail) + "}$")
if rhymes:
rhymes = unique(rhymes)
if word in rhymes:
rhymes.remove(word)
return rhymes
print("random general rhyme: tried all combos, didn't find anything!")
return []
def random_general_rhyme(word, phones=None, search_option="end"):
""" Return a list of rhymes where a random combination of phonemes match
The conditions for a general rhyme between words are:
(1) Any possible phonetic similarity between the final stressed vowel and
subsequent phonemes.
If phones argument not given, phones/pronunciation used will default to the
first in the list of phones returned for word. If no rhyme is found, an
empty list is returned.
:param word: a word
:param phones: specific CMUdict phonemes string for word (default None)
:param search_option option for regex search. (default "end")
:return: a list of rhymes for word, where specific rhyme is random
"""
if phones is None:
phones = first_phones_for_word(word)
if phones == "":
return []
else:
if phones not in pronouncing.phones_for_word(word):
raise ValueError(phones + " not phones for +" + word)
if not phones:
raise ValueError("phonemes string is empty")
rp = pronouncing.rhyming_part(phones)
search_combos = wildcard_mix_phones_regex_searches(rp)
while search_combos:
search = random.choice(search_combos)
if search_option == "end":
rhymes = pronouncing.search(search + "$")
elif search_option == "begin":
rhymes = pronouncing.search("^" + search)
elif search_option == "whole":
rhymes = pronouncing.search("^" + search + "$")
else:
raise ValueError("search_option should be 'end', 'begin', or 'whole'")
if rhymes:
rhymes = unique(rhymes)
if word in rhymes:
rhymes.remove(word)
return rhymes
else:
search_combos.remove(search)
print("random general rhyme: tried all combos, didn't find anything!")
return []
def random_match_phones(word, phones=None):
"""Returns words that match a random combination of phonemes
This is like a random general rhyme, however instead of just the
last syllable portion, it's the entire word.
:param word: word that should be in the CMU Pronouncing Dictionary
:param phones: specific phonemes to rhyme with (default None)
:return: a word that shares a random combinations of phonemes
"""
if phones is None:
phones = first_phones_for_word(word)
if phones == "":
return []
else:
if phones not in pronouncing.phones_for_word(word):
raise ValueError("phonemes and word don't match")
if not phones:
raise ValueError("phonemes string is empty")
search_list = wildcard_mix_phones_regex_searches(phones)
while search_list:
search = random.choice(search_list)
rhymes = pronouncing.search(search)
if rhymes:
rhymes = unique(rhymes)
if word in rhymes:
rhymes.remove(word)
return rhymes
else:
search_list.remove(search)
print("random general match phones: tried all combos, didn't find anything!")
return []
def assonance_slant_rhyme(word, phones=None):
""" Returns slant rhymes defined by assonance i.e. matching vowels.
The conditions for an assonance slant rhyme between words are:
(1) The last stressed vowel and subsequent phonemes match all vowels.
If phones argument not given, phones/pronunciation used will default to the
first in the list of phones returned for word. If no rhyme is found, an
empty list is returned.
Slant rhymes seems to have various different meanings. I went ahead and set
the condition of having to have ALL the same vowels and different
consonants or vice versa.
:param word: a word
:param phones: specific CMUdict phonemes string for word (default None)
:return: a list of assonance slant rhymes for word
"""
if phones is None:
phones = first_phones_for_word(word)
if phones == "":
return []
else:
if phones not in pronouncing.phones_for_word(word):
raise ValueError(phones + " not phones for +" + word)
if not phones:
raise ValueError("phonemes string is empty")
phones_list = phones.split()
search_list = []
for i in range(len(phones_list)-1, -1, -1):
phone = phones_list[i]
if check_if_non_stressed_vowel(phone):
search_list.append(phone[:2]+'.')
elif check_if_stressed_vowel(phone):
search_list.append(phone[:2]+'.') #ignore stress
search_list.reverse()
search = ' '.join(search_list)
rhymes = pronouncing.search(search + "$")
rhymes = unique(rhymes)
if word in rhymes:
rhymes.remove(word)
return rhymes
elif check_if_consonant(phone):
search_list.append('.{1,3}')
return []
def consonance_slant_rhyme(word, phones=None):
""" Returns slant rhymes defined by consonance i.e. matching consonants.
The conditions for a consonance slant rhyme between words are:
(1) The last stressed vowel and subsequent phonemes match all consonants.
If phones argument not given, phones/pronunciation used will default to the
first in the list of phones returned for word. If no rhyme is found, an
empty list is returned.
Slant rhymes seems to have various different meanings. I went ahead and set
the condition of having to have ALL the same vowels and different
consonants or vice versa.
:param word: a word
:param phones: specific CMUdict phonemes string for word (default None)
:return: a list consonance slant rhymes for word
"""
if phones is None:
phones = first_phones_for_word(word)
if phones == "":
return []
else:
if phones not in pronouncing.phones_for_word(word):
raise ValueError(phones + " not phones for +" + word)
if not phones:
raise ValueError("phonemes string is empty")
phones_list = phones.split()
search_list = []
for i in range(len(phones_list)-1, -1, -1):
phone = phones_list[i]
if check_if_stressed_vowel(phone):
search_list.append('.{1,3}')
if all_the_same(search_list, '.{1,3}') is True:
break
search_list.reverse()
search = ' '.join(search_list)
rhymes = pronouncing.search(search + "$")
rhymes = unique(rhymes)
if word in rhymes:
rhymes.remove(word)
return rhymes
elif check_if_non_stressed_vowel(phone):
search_list.append('.{1,3}')
elif check_if_consonant(phone):
search_list.append(phone)
return []
def assonance(word, phones=None, search_direction=None, match_limit=None):
"""Returns words that have assonance
:param word: word that should be in the CMU Pronouncing Dictionary
:param phones: specific phonemes to rhyme with (default None)
:param match_limit: limit number of words to return
:return: a word that has repition of vowel sounds to input word
"""
if phones is None:
phones = first_phones_for_word(word)
if phones == "":
return []
else:
if phones not in pronouncing.phones_for_word(word):
raise ValueError(phones + " not phones for +" + word)
if not phones:
raise ValueError("phonemes string is empty")
phones_list = phones.split()
if search_direction == "backward":
phones_list.reverse()
search_list = []
match_cnt = 0
for phone in phones_list:
if check_if_consonant(phone):
search_list.append('.')
elif check_if_vowel(phone):
search_list.append(phone)
match_cnt += 1
if (match_limit is not None) and (match_cnt == match_limit):
break
if search_direction == "backward":
search = ' '.join(search_list.reverse()) + "$"
elif search_direction == "forward":
search = "^" + ' '.join(search_list)
else:
search = ' '.join(search_list)
rhymes = pronouncing.search(search)
rhymes = unique(rhymes)
if word in rhymes:
rhymes.remove(word)
return rhymes
def consonance(word, phones=None, search_direction=None, match_limit=None):
"""Returns words that have consonance
:param word: word that should be in the CMU Pronouncing Dictionary
:param phones: specific phonemes to rhyme with (default None)
:param match_limit: limit number of words to return
:return: a word that has repition of consonance sounds to input word
"""
if phones is None:
phones = first_phones_for_word(word)
if phones == "":
return []
else:
if phones not in pronouncing.phones_for_word(word):
raise ValueError("phonemes and word don't match")
if not phones:
raise ValueError("phonemes string is empty")
phones_list = phones.split()
if search_direction == "backward":
phones_list.reverse()
search_list = []
match_cnt = 0
for phone in phones_list:
if check_if_vowel(phone):
search_list.append('.{1,3}')
elif check_if_consonant(phone):
search_list.append(phone)
match_cnt += 1
if (match_limit is not None) and (match_cnt == match_limit):
break
if search_direction == "backward":
search = ' '.join(search_list.reverse()) + "$"
elif search_direction == "forward":
search = "^" + ' '.join(search_list)
else:
search = ' '.join(search_list)
rhymes = pronouncing.search(search)
rhymes = unique(rhymes)
if word in rhymes:
rhymes.remove(word)
return rhymes
def alliteration(word):
return consonance(word, "forward", 1)
def wildcard_mix_phones_regex_searches(phones, stress=False):
"""Generates all combinations of regex strings where phoneme in 'phones' is a wildcard ('.')
e.g. ['HH IY1 R'],['HH IY1 .{1,3}'],['HH .{1,3} R'],
['.{1,3} IY1 R'], ...['.{1,3} .{1,3} .{1,3}']
:param phones: CMU Pronouncing Dictionary phonemes string
:param stress: if stress portion of vowel is included or nont (a '.')
:return: list of regex search strings where phonemes replaced with wildcard
"""
phones_list = phones.split()
product_factors = []
for phone in phones_list:
flist = ['.{1,3}']
if stress is False and check_if_vowel(phone):
flist.append(phone[:2]+'.') # ignore stress
else:
flist.append(phone)
product_factors.append(flist)
combos = list(itertools.product(*product_factors))
combos.remove(combos[0]) # should be case where ['.', '.', ... '.']
search_combos = [' '.join(list(item)) for item in combos]
return search_combos
def rhyme_same_stress(word):
timeout_timer = 0
# print('in the stress loop')
while(True):
phones = pronouncing.phones_for_word(word)
phone = random.choice(phones)
word_stress = pronouncing.stresses(phone)
rhyme = rhyme_type_random(word)
phones = pronouncing.phones_for_word(rhyme)
for phone in phones:
rhyme_stress = pronouncing.stresses(phone)
if word_stress == rhyme_stress:
return rhyme
print(timeout_timer)
if timeout_timer == 10:
return rhyme
timeout_timer += 1
def sim_word_for_phones(phones, word_list=[], sim_perc = 0.25):
"""Finds a word that has shared phonemes to input phonemes
:param phones: CMU Pronouncing Dictionary phonemes string
:param word_list: list of words to limit search to
:sim_perc: threshold for similarity between phones and words in words_list
:return: word that has number of phonemes the same as phones
"""
search_combos = wildcard_mix_phones_regex_searches(phones)
random.shuffle(search_combos)
for sch in search_combos:
sch_list = sch.split(" ")
if sch_list.count(".{1,3}") < (1-sim_perc)*len(sch_list):
matches = pronouncing.search("^" + sch + "$")
if matches:
matches = unique(matches)
random.shuffle(matches)
for m in matches:
if word_list:
if m in word_list:
return(m)
else:
return(m)
return None
def rhyme_type_random(word):
rhyme_types = ['perfect', 'identical', 'random_match_phones', 'random_general',
'assonance', 'consonance', 'slant_assonance', 'slant_consonance']
rhymes = []
while rhyme_types:
rt = random.choice(rhyme_types)
if rt == 'perfect':
rhymes = perfect_rhyme(word)
elif rt == 'identical':
rhymes = identical_rhyme(word)
elif rt == 'random_match_phones':
rhymes = random_match_phones(word)
elif rt == 'random_general':
rhymes = random_general_rhyme(word)
elif rt == 'assonance':
rhymes = assonance(word)
elif rt == 'consonance':
rhymes = consonance(word)
elif rt == 'slant_assonance':
rhymes = assonance_slant_rhyme(word)
elif rt == 'slant_consonance':
rhymes = consonance_slant_rhyme(word)
if rhymes:
a_rhyme = random.choice(rhymes)
return a_rhyme
else:
rhyme_types.remove(rt)
return []