English MFA dictionary v2.0.0a#
@techreport{mfa_english_mfa_dictionary_2022,
author={McAuliffe, Michael and Sonderegger, Morgan},
title={English MFA dictionary v2.0.0a},
address={\url{https://mfa-models.readthedocs.io/pronunciation dictionary/English/English MFA dictionary v2_0_0a.html}},
year={2022},
month={May},
}
Acoustic models |
Installation#
Install from the MFA command line:
mfa model download dictionary english_mfa
Or download from the release page.
The dictionary available from the release page and command line installation has pronunciation and silence probabilities estimated as part acoustic model training (see Silence probability format and training pronunciation probabilities for more information. If you would like to use the version of this dictionary without probabilities, please see the plain dictionary.
Intended use#
This dictionary is intended for forced alignment of English transcripts.
This dictionary uses the MFA phone set for English, and was used in training the English MFA acoustic model. Pronunciations can be added on top of the dictionary, as long as no additional phones are introduced.
Performance Factors#
When trying to get better alignment accuracy, adding pronunciations is generally helpful, especially for different styles and dialects. The most impactful improvements will generally be seen when adding reduced variants that involve deleting segments/syllables common in spontaneous speech. Alignment must include all phones specified in the pronunciation of a word, and each phone has a minimum duration (by default 10ms). If a speaker pronounces a multisyllabic word with just a single syllable, it can be hard for MFA to fit all the segments in, so it will lead to alignment errors on adjacent words as well.
Ethical considerations#
Deploying any Speech-to-Text model into any production setting has ethical implications. You should consider these implications before use.
Demographic Bias#
You should assume every machine learning model has demographic bias unless proven otherwise. For pronunciation dictionaries, it is often the case that transcription accuracy and lexicon coverage for the prestige variety modeled in this dictionary compared to other variants. If you are using this dictionary in production, you should acknowledge this as a potential issue.
IPA Charts#
Consonants#
Obstruent symbols to the left of are unvoiced and those to the right are voiced.
Manner |
Labial |
Labiodental |
Dental |
Alveolar |
Alveopalatal |
Palatal |
Velar |
Glottal |
---|---|---|---|---|---|---|---|---|
Nasal |
Occurrences: 38,529 Examples: * toman: [t ʰ o w m n ̩] * mirth: [m ɝ θ] * miele: [m ʲ i ː ɫ] * mewks: [m ʲ ʉ ː k s] Occurrences: 4,166 Examples: * medea: [m ʲ i d ʲ i ː a] * mimi: [m ɪ m ʲ i] * pumy: [p ʲ ʉ ː m ʲ i] * meme: [m ʲ i ː m] Occurrences: 2,656 Examples: * hokum: [h o w k m ̩] * oram: [ɒ ː m ̩] * sodom: [s ɑ ɾ m ̩] * chasm: [c ʰ æ z m ̩] |
Occurrences: 2,267 Examples: * info: [ɪ ɱ f o w] * infix: [ɪ ɱ f ɪ k s] * oomph: [ʉ ɱ f] * comfy: [k ʰ ɐ ɱ f ʲ i] |
Occurrences: 67,052 Examples: * neg: [n ɛ ɡ] * nisan: [ɲ i ː s ɑ ː n] * conda: [k ʰ ɑ n d ə] * abuna: [ɑ b ʲ ʉ n ɑ] Occurrences: 7,480 Examples: * coven: [k ʰ ɐ v n ̩] * eoan: [i o w n ̩] * ryan: [ɹ a j n ̩] * luton: [l ʉ ː ʔ n ̩] |
Occurrences: 16,157 Examples: * napa: [ɲ æ p ə] * napo: [ɲ e j p ʰ ə w] * nabk: [ɲ æ b ə k] * inyo: [ɪ ɲ ə w] |
Occurrences: 14,369 Examples: * yonks: [j ɒ ŋ k s] * angus: [æ ŋ ɡ ə s] * tonks: [t ʰ ɑ ŋ k s] * prong: [p ɹ ɑ ŋ] |
|||
Stop |
Occurrences: 21,457 Examples: * ayelp: [ə j ɛ ɫ p] * jumpy: [d ʒ ɐ m p ʲ i] * posse: [p ʰ ɑ s i] * cups: [k ʰ ɐ p s] Occurrences: 27,452 Examples: * betsy: [b ɛ t s i] * brief: [b ɹ i ː f] * keble: [c ʰ i ː b ɫ ̩] * brien: [b ɹ a j n ̩] |
Occurrences: 108 Examples: * both: [b o w t ̪] * thing: [t ̪ ɪ ŋ] * width: [w ɪ t ̪] * thumb: [t ̪ ɐ m] Occurrences: 224 Examples: * with: [w ɪ d ̪] * they: [d ̪ e j] * thems: [d ̪ m ̩ z] * them: [d ̪ m ̩] |
Occurrences: 55,379 Examples: * ranch: [ɹ æ n t ʃ] * peta: [p ʰ i ː t ə] * shot: [ʃ ɑ t] * fust: [f ɐ s t] Occurrences: 42,445 Examples: * undo: [ɐ n d ʉ ː] * devas: [d ɛ v ə s] * egad: [i ː ɡ æ d] * jie: [d ʒ i] |
Occurrences: 9,946 Examples: * honky: [ɒ ː ɲ c i] * creek: [c ɹ i ː k] * cack: [c ʰ æ k] * kilim: [c ʰ i ʎ i m] Occurrences: 5,970 Examples: * gift: [ɟ ɪ f t] * greet: [ɟ ɹ i ː t] * gland: [ɟ ʎ æ n d] * game: [ɟ e j m] |
Occurrences: 33,321 Examples: * kamba: [k ʰ ɑ ː m b ə] * mac: [m æ k] * cubby: [k ʰ ɐ b ʲ i] * marc: [m ɑ ɹ k] Occurrences: 14,178 Examples: * gov: [ɡ ɐ v] * regal: [ɹ i ː ɡ ɫ ̩] * gown: [ɡ a w n] * cig: [s ɪ ɡ] |
Occurrences: 6,730 Examples: * mute: [m ʲ ʉ ː ʔ] * orta: [ɒ ː ʔ ə] * lit: [ʎ ɪ ʔ] * butte: [b ʲ ʉ ː ʔ] |
||
Affricate |
Occurrences: 8,697 Examples: * chopp: [t ʃ ɑ p] * chon: [t ʃ ɑ n] * welch: [w ɛ ɫ t ʃ] * coach: [k ʰ o w t ʃ] Occurrences: 11,671 Examples: * fiji: [f ʲ i d ʒ i] * jon: [d ʒ ɑ n] * cabg: [c ʰ æ b ɪ d ʒ] * josh: [d ʒ ɑ ʃ] |
|||||||
Sibilant |
Occurrences: 79,648 Examples: * canis: [c ʰ æ ɲ ɪ s] * dixon: [d ɪ k s ə n] * butsu: [b ʊ t s ʉ] * vases: [v æ s ɪ z] Occurrences: 31,451 Examples: * sines: [s a j n z] * pews: [p ʲ ʉ ː z] * jabs: [d ʒ æ b z] * coils: [k ʰ ɔ j ɫ z] |
Occurrences: 15,280 Examples: * tule: [t ʃ ʉ ː ʎ i] * etude: [e j t ʃ ʉ ː d] * stash: [s t æ ʃ] * who: [d ɐ b ɫ ̩ j ʉ ː e j t ʃ o w] Occurrences: 1,226 Examples: * jut: [d ʒ ɐ t] * regex: [ɹ ɛ d ʒ ɛ k s] * jokes: [d ʒ o w k s] * dig: [d ɪ d ʒ] |
||||||
Fricative |
Occurrences: 20,600 Examples: * forbs: [f ɒ ː b z] * golf: [ɡ ɑ ɫ f] * fall: [f ɒ ɫ] * foe: [f o w] Occurrences: 2,615 Examples: * feuds: [f ʲ ʉ ː d z] * fink: [f ʲ i ŋ k] * fusel: [f ʲ ʉ ː z ɫ ̩] * rafi: [ɹ ɑ ː f ʲ i] Occurrences: 14,770 Examples: * garve: [ɡ ɑ ː v] * ogive: [o w d ʒ a j v] * via: [v a j ə] * evert: [i v ɝ t] Occurrences: 1,708 Examples: * uvula: [j ʉ ː v ʲ ʊ l ə] * civvy: [s ɪ v ʲ i] * vig: [v ʲ i ɡ] * civey: [s ɪ v ʲ i] |
Occurrences: 5,746 Examples: * werth: [w ɜ ː θ] * garth: [ɡ ɑ ɹ θ] * thunk: [θ ɐ ŋ k] * theo: [θ i o w] Occurrences: 1,326 Examples: * thus: [ð ɐ s] * rathe: [ɹ e j ð] * kithe: [c ʰ a j ð] * paths: [p ʰ æ ð z] |
Occurrences: 2,104 Examples: * haem: [ç i ː m] * hib: [ç ɪ b] * haydn: [ç e j ɾ ə n] * hitt: [ç i t] |
Occurrences: 8,413 Examples: * haxor: [h æ k s ɒ ɹ] * het: [h ɛ t] * hilum: [h a j l m ̩] * homey: [h o w m ʲ i] |
||||
Approximant |
Occurrences: 12,066 Examples: * cmos: [s i ː ɛ m ə w z] * tasso: [t ʰ æ s o w] * bald: [b ə w ɫ d] * wears: [w ɛ ɹ z] |
Occurrences: 70,152 Examples: * fro: [f ɹ o w] * frosh: [f ɹ ɒ ʃ] * resto: [ɹ ɛ s t ə w] * orse: [ɒ ɹ s] |
Occurrences: 3,260 Examples: * virid: [v a j ɹ ɪ d] * yet: [j ɛ t] * dire: [d a j ɹ] * hired: [h a j ɹ d] |
|||||
Tap |
Occurrences: 10,267 Examples: * fatal: [f e j ɾ ɫ ̩] * blady: [b l e j ɾ i] * roti: [ɹ o w ɾ i] * judah: [d ʒ ʉ ː ɾ ə] |
|||||||
Lateral |
Occurrences: 28,850 Examples: * tcler: [t ʰ ɪ k l ɚ] * lundy: [l ɐ n d ʲ i] * layin: [l e j ɪ n] * plots: [p l ɑ t s] Occurrences: 16,542 Examples: * alecy: [e j ɫ s i] * sele: [s i ː ɫ] * hoyle: [h ɔ j ɫ ̩] * bebel: [b e j b ɫ ̩] Occurrences: 8,114 Examples: * cupel: [c ʰ ʉ ː p ɫ ̩] * ewell: [j ʉ ː ɫ ̩] * duval: [d ʲ ʉ ː v ɫ ̩] * idle: [a j ɾ ɫ ̩] |
Occurrences: 20,139 Examples: * delia: [d ʲ i ː ʎ ɪ ə] * clip: [c ʎ ɪ p] * luau: [ʎ ʉ a w] * marly: [m ɑ ɹ ʎ i] |
Vowels#
Vowel symbols to the left of are unrounded and those to the right are rounded.
Front |
Near-Front |
Central |
Near-Back |
Back |
|
---|---|---|---|---|---|
Close |
Occurrences: 62,307 Examples: * okdk: [ə w c ɪ d ə w c i] * megan: [m ʲ i ː ɡ n ̩] * tariq: [t ʰ a ɹ i k] * kiel: [c ʰ i ɫ] Occurrences: 17,759 Examples: * fetus: [f ʲ i ː ɾ ə s] * jine: [d ʒ i ː n] * beads: [b ʲ i ː d z] * aacdp: [e j e j s i ː d ʲ i ː p ʰ i ː] |
Occurrences: 2,910 Examples: * poops: [p ʰ ʉ ː p s] * ensue: [ɪ n s ʉ ː] * bute: [b ʲ ʉ ː t] * hew: [ç ʉ ː] Occurrences: 9,006 Examples: * ousia: [ʉ ː s ɪ ə] * glute: [ɟ ʎ ʉ ː t] * kuru: [k ʰ ʊ ɹ ʉ ː] * susu: [s ɐ s ʉ ː] |
Occurrences: 3,895 Examples: * nuer: [n u ː a] * roofs: [ɹ u ː f s] * bubo: [b ʲ u ː b o] * poof: [p ʰ u ː f] Occurrences: 4,365 Examples: * hoot: [h u ː t] * lues: [l u ː i ː z] * rus: [ɹ u ː s] * tuvan: [t ʃ u ː v a n] |
||
Occurrences: 73,922 Examples: * whish: [w ɪ ʃ] * jakin: [d ʒ e j c ɪ n] * muniz: [m ʲ ʉ ː ɲ ɪ t s] * muses: [m ʲ ʉ ː z ɪ z] |
Occurrences: 7,940 Examples: * huron: [ç ʊ ɹ ɑ n] * fully: [f ʊ ʎ i] * pula: [p ʰ ʊ l ə] * wud: [w ʊ d] |
||||
Close-Mid |
Occurrences: 6,865 Examples: * raid: [ɹ e j d] * deify: [d e j ə f a j] * grego: [ɟ ɹ e j ɡ ə w] * made: [m e j d] Occurrences: 17,938 Examples: * ame: [e j m ʲ i] * waves: [w e j v z] * jays: [d ʒ e j z] * rays: [ɹ e j z] |
Occurrences: 6,444 Examples: * junco: [d ʒ ɐ ŋ k o w] * cello: [s ɛ l o w] * rhoda: [ɹ o w ɾ ə] * polk: [p ʰ o w ɫ k] Occurrences: 10,138 Examples: * close: [k l o w s] * tyro: [t ʰ a j ɹ o w] * oding: [o w d ʲ i ː ɪ ŋ] * vitro: [v ʲ i ː t ɹ o w] |
|||
Occurrences: 71,603 Examples: * porto: [p ʰ ɒ ː t ʰ ə w] * grope: [ɡ ɹ ə w p] * allen: [æ l ə n] * sligo: [s l a j ɡ ə w] Occurrences: 9,455 Examples: * solar: [s o w l ɚ] * armor: [ɑ ɹ m ɚ] * sider: [s a j ɾ ɚ] * oder: [o w ɾ ɚ] |
|||||
Open-Mid |
Occurrences: 43,024 Examples: * cares: [c ʰ ɛ ɹ z] * egdon: [ɛ ɡ d ə n] * ucla: [j ʉ s i ɛ l e j] * lnav: [ɛ l ɲ e j v] Occurrences: 1,534 Examples: * verde: [v ɛ ː d e j] * fared: [f ɛ ː d] * layer: [l ɛ ː] * vair: [v ɛ ː] |
Occurrences: 312 Examples: * turku: [t ʰ ɜ k ʉ ː] * irvin: [ɜ ː v ɪ n] * urn: [ɜ ː n] * curse: [k ʰ ɜ ː s] Occurrences: 3,347 Examples: * skirt: [s k ɜ ː t] * lure: [l ɜ ː] * irma: [ɜ ː m ə] * yern: [j ɜ ː n] Occurrences: 3,483 Examples: * merc: [m ɝ k] * first: [f ɝ s t] * burrs: [b ɝ z] * erles: [ɝ ɫ z] |
Occurrences: 21,365 Examples: * ducky: [d ɔ c ʰ i] * joins: [d ʒ ɔ j n z] * gaw: [ɡ ɔ] * savoy: [s ə v ɔ j] |
||
Occurrences: 25,228 Examples: * rash: [ɹ æ ʃ] * brans: [b ɹ æ n z] * fangs: [f æ ŋ z] * tampa: [t ʰ æ m p ə] |
Occurrences: 11,030 Examples: * cups: [k ʰ ɐ p s] * m'lud: [m ɪ l ɐ d] * sucky: [s ɐ c i] * unget: [ɐ ɲ ɟ ɛ t] |
||||
Open |
Occurrences: 32,215 Examples: * nigh: [n a j] * lisa: [l a j z ə] * rheas: [ɹ i ː a z] * there: [d ̪ e a] Occurrences: 1,879 Examples: * quirt: [k w a ː t] * blurs: [b l a ː z] * loess: [l a ː s] * dern: [d a ː n] |
Occurrences: 11,332 Examples: * hock: [h ɑ k] * darts: [d ɑ ɹ t s] * darla: [d ɑ ɹ l ə] * task: [t ʰ ɑ ː s k] Occurrences: 6,813 Examples: * laos: [l ɑ ː o w s] * parse: [p ʰ ɑ ː z] * drama: [d ɹ ɑ ː m ə] * yamen: [j ɑ ː m n ̩] Occurrences: 15,695 Examples: * rioja: [ɹ i ː ɒ k ə] * scots: [s k ɒ t s] * laude: [l ɒ ː d] * skald: [s k ɒ ː ɫ d] Occurrences: 5,974 Examples: * acorn: [e j k ʰ ɒ ː n] * daub: [d ɒ ː b] * chaw: [t ʃ ɒ ː] * caut: [k ʰ ɒ ː t] |
Diphthongs#
aj
aw
ej
ow
ɔj
əw