English MFA dictionary v3.0.0#
@techreport{mfa_english_mfa_dictionary_2024,
author={McAuliffe, Michael and Sonderegger, Morgan},
title={English MFA dictionary v3.0.0},
address={\url{https://mfa-models.readthedocs.io/pronunciation dictionary/English/English MFA dictionary v3_0_0.html}},
year={2024},
month={Feb},
}
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](https://raw.githubusercontent.com/MontrealCorpusTools/mfa-models/main/dictionary/english/mfa/English MFA dictionary v3_0_0.dict).
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: 39,803 Examples: * trim: [tʲ ɹ ɪ m] * ament: [ej m ɛ n t] * lambs: [l a m z] * gambo: [ɡ a m b o] Occurrences: 8,153 Examples: * meads: [mʲ iː d z] * omega: [əw mʲ iː ɡ ə] * midst: [mʲ i d s t] * meets: [mʲ iː t s] Occurrences: 8 Examples: * poems: [pʰ ow m̩ z] * album: [æ ɫ b m̩] * him: [m̩] * some: [s m̩] |
Occurrences: 1 Examples: |
Occurrences: 77,947 Examples: * burns: [b ɜː n z] * varna: [v aː n a] * vines: [v aj n z] * namco: [n a m kʰ əw] Occurrences: 98 Examples: * giant: [dʒ aj n̩ t] * often: [ɒ f t n̩] * open: [ow p n̩] * isn't: [ɪ z n̩] |
Occurrences: 15,840 Examples: * nadir: [ɲ ej dʲ ɪ ə] * news: [ɲ uː z] * nap: [ɲ æ p] * snail: [s ɲ ej ɫ] |
Occurrences: 15,493 Examples: * mango: [m æ ŋ ɡ ow] * shank: [ʃ a ŋ k] * longa: [l ɒ ŋ ɡ ə] * inked: [ɪ ŋ k t] |
|||
Stop Plain |
Occurrences: 20,997 Examples: * spark: [s p ɑ ɹ k] * sloop: [s l ʉː p] * hap: [h æ p] * camps: [kʰ a m p s] Occurrences: 3,212 Examples: * puma: [pʲ uː m a] * udupi: [j ʉː d ʉː pʲ i] * aspis: [æ s pʲ ɪ s] * tuppy: [tʰ ɐ pʲ i] Occurrences: 29 Examples: * puy: [pʷ iː] Occurrences: 25,552 Examples: * balk: [b ɒː ɫ k] * bout: [b aw t] * taboo: [tʰ a b ʉː] * bart: [b ɑ ɹ t] Occurrences: 5,316 Examples: * bilge: [bʲ ɪ ɫ dʒ] * benet: [bʲ ɪ n ɛ t] * robyn: [ɹ əw bʲ ɪ n] * sibyl: [s ɪ bʲ ɪ ɫ] |
Occurrences: 130 Examples: * youth: [j ʉː t̪] * worth: [w ɔ t̪] * both: [b əw t̪] * thin: [t̪ ɪ n] Occurrences: 115 Examples: * than: [d̪ æ n] * these: [d̪ iː z] * this: [d̪ i s] * that: [d̪ æ t] |
Occurrences: 52,430 Examples: * bigot: [bʲ ɪ ɡ ə t] * altar: [ɔ l t a] * wants: [w ə n t s] * bouts: [b aw t s] Occurrences: 19,256 Examples: * putin: [pʰ uː tʲ i n] * trio: [tʲ ɹ i əw] * team: [tʲ iː m] * steen: [s tʲ iː n] Occurrences: 267 Examples: * tweet: [tʷ iː t] * twill: [tʷ ɪ ɫ] * twigs: [tʷ ɪ ɡ z] * tweak: [tʷ iː k] Occurrences: 39,706 Examples: * iliad: [ɪ ʎ i a d] * jihad: [dʒ ə h ɑː d] * dari: [d ɑː ɹ i] * lands: [l a n d z] Occurrences: 12,826 Examples: * lydia: [ʎ i dʲ i a] * deer: [dʲ ɪ ə] * today: [tʰ ɔ dʲ e] * idiot: [ɪ dʲ i ə t] |
Occurrences: 9,105 Examples: * crape: [c ɹ ej p] * crips: [c ɹ ɪ p s] * tiki: [tʲ iː c i] * wiske: [w ɪ s c i] Occurrences: 977 Examples: * squad: [s cʷ æ d] * queer: [cʷ i a] * quill: [cʷ i ɫ] * quell: [cʷ ɛ ɫ] Occurrences: 4,529 Examples: * gated: [ɟ ej tʲ ɪ d] * gray: [ɟ ɹ ej] * saggy: [s æ ɟ i] * grace: [ɟ ɹ ej s] Occurrences: 202 Examples: * gwen: [ɟʷ ɛ n] * gwynn: [ɟʷ ɪ n] * guido: [ɟʷ i d ow] * gwin: [ɟʷ ɪ n] |
Occurrences: 32,577 Examples: * mimic: [mʲ ɪ mʲ ɪ k] * cox: [kʰ ɔ k s] * pock: [pʰ ɒ k] * skuas: [s k ʉː ə z] Occurrences: 1,442 Examples: * kwok: [kʷ ɔ k] * quark: [kʷ ɒ ɹ k] * quart: [kʷ ɒ ɹ t] * croix: [kʷ ɒː] Occurrences: 17,075 Examples: * going: [ɡ ow ɪ ŋ] * gael: [ɡ ej ɫ] * eagle: [iː ɡ ə ɫ] * brag: [b ɹ æ ɡ] Occurrences: 163 Examples: * guava: [ɡʷ ɑ v ə] * guano: [ɡʷ a n o] * agua: [ɑ ɡʷ ə] * guang: [ɡʷ ɑː ŋ] |
Occurrences: 102 Examples: * what: [w ɒ ʔ] * putin: [pʲ ʉː ʔ ə n] * got: [ɡ ɑ ʔ] |
||
Aspirated |
Occurrences: 14,454 Examples: * ports: [pʰ ɔ t s] * agape: [ɑ ɡ ɑː pʰ ej] * paula: [pʰ ɒː l ə] * peak: [pʰ iː k] |
Occurrences: 14,838 Examples: * text: [tʰ ɛ k s t] * today: [tʰ ʊ d ej] * tulsi: [tʰ ʊ ɫ s iː] * tahoe: [tʰ ɑː h ow] |
Occurrences: 5,987 Examples: * caged: [cʰ ej dʒ d] * kit: [cʰ i t] * kea: [cʰ iː ə] * nokia: [n ɔ cʰ i a] |
Occurrences: 15,859 Examples: * cosh: [kʰ ɒ ʃ] * narco: [n ɑː kʰ əw] * kern: [kʰ aː n] * cuddy: [kʰ ɔ dʲ i] |
||||
Affricate |
Occurrences: 7,320 Examples: * mitch: [mʲ i tʃ] * char: [tʃ ɑː] * two: [tʃ uː] * roch: [ɹ ɒ tʃ] Occurrences: 11,779 Examples: * saj: [s ɑː dʒ] * jeet: [dʒ i t] * july: [dʒ ʉː l aj] * jimi: [dʒ ɪ mʲ iː] |
|||||||
Sibilant |
Occurrences: 81,280 Examples: * stay: [s t ej] * hans: [h ɑː n s] * luce: [l ʉː s] * soc: [s əw ʃ] Occurrences: 33,119 Examples: * axons: [æ k s ɑ n z] * nikes: [n aj c i z] * arabs: [ej ɹ æ b z] * laser: [l ej z ɚ] |
Occurrences: 16,190 Examples: * xiang: [ʃ ɑː ŋ] * ocean: [əw ʃ ə n] * shush: [ʃ ɔ ʃ] * niche: [ɲ i ʃ] Occurrences: 896 Examples: * luge: [ʎ ʉː ʒ] * asian: [ej ʒ ə n] * zhang: [ʒ æ ŋ] * deja: [d ej ʒ ə] |
||||||
Fricative |
Occurrences: 18,515 Examples: * frets: [f ɹ ɛ t s] * frist: [f ɹ ɪ s t] * morph: [m ɔ f] * fare: [f ɛː] Occurrences: 5,050 Examples: * fit: [fʲ ɪ t] * comfy: [kʰ ɔ m fʲ i] * philo: [fʲ ɪ l ow] * filet: [fʲ i ʎ i t] Occurrences: 3 Examples: Occurrences: 13,845 Examples: * vert: [v ɝ t] * lavra: [l a v ɹ ə] * venn: [v ɛ n] * wurst: [v ʊ ɹ s t] Occurrences: 3,127 Examples: * veers: [vʲ ɪ ɹ z] * ovid: [ɑ vʲ ɪ d] * mavis: [m e vʲ i s] * ivy: [aj vʲ i] Occurrences: 8 Examples: |
Occurrences: 5,841 Examples: * cloth: [k l ɔ θ] * earth: [aː θ] * tithi: [tʲ iː θ i] * birth: [b ɝ θ] Occurrences: 1,364 Examples: * paths: [pʰ æ ð z] * dhabi: [ð ɑː bʲ i] * the: [ð iː] * that: [ð æ t] |
Occurrences: 2,154 Examples: * heme: [ç iː m] * hymn: [ç i m] * hiss: [ç ɪ s] * hewn: [ç uː n] |
Occurrences: 8,526 Examples: * hansa: [h a n s ə] * has: [h æ s] * hired: [h aj ə d] * hajj: [h ɔ dʒ] |
||||
Approximant |
Occurrences: 9,186 Examples: * waves: [w ej v z] * wire: [w aj ə] * wings: [w ɪ n z] * woman: [w ʊ m a n] |
Occurrences: 70,108 Examples: * prams: [p ɹ a m z] * drava: [d ɹ ɑː v ə] * prod: [p ɹ ɒ d] * prior: [p ɹ aj ə] |
Occurrences: 3,285 Examples: * yargo: [j ɑ ɹ ɡ ow] * yowie: [j aw w i] * utica: [j uː tʲ i k a] * yoke: [j o k] |
|||||
Tap |
Occurrences: 196 Examples: * gad: [ɡ æ ɾ] * metal: [m ɛ ɾ ə ɫ] * head: [h ɛ ɾ] * it'll: [ɪ ɾ ɫ̩] Occurrences: 97 Examples: * duty: [dʲ ʉː ɾʲ i] * body: [b ɔ ɾʲ i] * ready: [ɹ ɛ ɾʲ i] * video: [vʲ i ɾʲ i o] Occurrences: 35 Examples: * many: [m ɪ ɾ̃ i] * under: [ɐ ɾ̃ ɚ] * tiny: [tʰ aj ɾ̃ i] |
|||||||
Lateral |
Occurrences: 29,417 Examples: * loire: [l w ɑ ɹ] * lukas: [l ʉ k ə z] * tyler: [tʰ aj l ɚ] * flubs: [f l ɔ b z] Occurrences: 24,671 Examples: * helps: [h ɛ ɫ p s] * bible: [b aj b ə ɫ] * tile: [tʰ aj ɫ] * colt: [kʰ ɒ ɫ t] Occurrences: 158 Examples: * total: [tʰ əw t ɫ̩] * vital: [v aj t ɫ̩] * evil: [iː v ɫ̩] * civil: [s ɪ v ɫ̩] |
Occurrences: 20,823 Examples: * aliyu: [a ʎ i j uː] * limit: [ʎ ɪ mʲ ɪ t] * lil: [ʎ i ɫ] * hally: [h ɔ 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: 63,751 Examples: * reset: [ɹ i s ɛ t] * tasty: [tʰ ej s tʲ i] * piggy: [pʰ ɪ ɟ i] * raspy: [ɹ a s pʲ i] Occurrences: 17,066 Examples: * gaea: [dʒ iː ə] * weak: [w iː k] * dion: [dʲ iː ɑ n] * davie: [d ej vʲ iː] |
Occurrences: 2,833 Examples: * moody: [m ʉ dʲ i] * usa: [j ʉ ɛ s ej] * luis: [ʎ ʉ ɪ s] * zeus: [z ʉ s] Occurrences: 8,227 Examples: * cued: [cʰ ʉː d] * prune: [p ɹ ʉː n] * truth: [tʃ ɹ ʉː θ] * unite: [j ʉː n aj t] |
Occurrences: 4,035 Examples: * bacon: [b e k u n] * upton: [ɔ p t u n] * ovum: [o v u m] * durum: [dʲ uː u m] Occurrences: 4,667 Examples: * fuels: [fʲ uː ɫ z] * argue: [a ɡ uː] * muti: [m uː tʰ i] * boop: [b uː p] |
||
Occurrences: 72,797 Examples: * rivet: [ɹ ɪ v ə t] * cabin: [cʰ æ bʲ ɪ n] * sin: [s ɪ n] * wick: [w ɪ k] |
Occurrences: 8,064 Examples: * cycle: [s aj k ʊ ɫ] * stook: [s t ʊ k] * buren: [b ʊ ɹ ə n] * cure: [cʰ ʊ a] |
||||
Close-Mid |
Occurrences: 7,679 Examples: * blaby: [b l e bʲ i] * bay: [b e] * navel: [n e v ɛ ɫ] * rates: [ɹ e t s] Occurrences: 17,233 Examples: * raven: [ɹ ej v ə n] * malay: [m ej l ej] * ages: [ej dʒ ɪ z] * pace: [pʰ ej s] |
Occurrences: 7,012 Examples: * prove: [p ɹ o v] * loess: [l o i s] * poll: [pʰ o ɫ] * volga: [v o ɫ ɡ a] Occurrences: 10,116 Examples: * ago: [ə ɡ ow] * mole: [m ow l ej] * roca: [ɹ ow k ə] * poems: [pʰ ow m̩ z] |
|||
Occurrences: 86,312 Examples: * fanon: [f a n ə n] * bison: [b aj s ə n] * haydn: [ç ej d ə n] * alma: [a ɫ m ə] Occurrences: 9,826 Examples: * luger: [ʎ ʉ ɡ ɚ] * caper: [cʰ ej p ɚ] * acre: [ej k ɚ] * alder: [ɒː ɫ d ɚ] |
|||||
Open-Mid |
Occurrences: 46,399 Examples: * caron: [cʰ ɛ ɹ ə n] * etna: [ɛ t n ə] * selma: [s ɛ ɫ m a] * xtra: [ɛ k s t ɹ ə] Occurrences: 1,561 Examples: * dares: [d ɛː z] * spare: [s p ɛː] * flare: [f l ɛː] * carew: [cʰ ɛː ɹ ʉː] |
Occurrences: 310 Examples: * viera: [vʲ iː ɜ ɹ a] * bury: [b ɜ ɹ i] * circe: [s ɜ s i] * burj: [b ɜ dʒ] Occurrences: 3,184 Examples: * birk: [b ɜː k] * first: [f ɜː s t] * worth: [w ɜː θ] * ernie: [ɜː ɲ i] Occurrences: 3,489 Examples: * third: [s ɝ d] * burg: [b ɝ ɡ] * word: [w ɝ d] * erne: [ɝ n] |
Occurrences: 23,536 Examples: * spall: [s p ɔ ɫ] * malta: [m ɔ ɫ t a] * bunks: [b ɔ ŋ k s] * tor: [tʰ ɔ] |
||
Occurrences: 15,920 Examples: * altos: [æ ɫ t ow z] * anwar: [æ n w ɚ] * gavin: [ɡ æ vʲ ɪ n] * paths: [pʰ æ θ s] |
Occurrences: 10,494 Examples: * jumbo: [dʒ ɐ m b ow] * upper: [ɐ p ɚ] * moray: [m ɐ ɹ i] * tump: [tʰ ɐ m p] |
||||
Open |
Occurrences: 47,477 Examples: * exact: [ɪ ɡ z a k t] * trash: [t ɹ a ʃ] * maga: [m a ɡ ə] * value: [v a ʎ ʉː] Occurrences: 2,141 Examples: * sperm: [s p aː m] * light: [l aː t] * hern: [h aː n] * price: [p ɹ aː s] |
Occurrences: 11,104 Examples: * togs: [tʰ ɑ ɡ z] * dante: [d ɑ n tʰ ej] * lat: [l ɑ t] * agog: [ə ɡ ɑ ɡ] Occurrences: 7,128 Examples: * bah: [b ɑː] * barth: [b ɑː θ] * bara: [b ɑː ɹ ə] * noir: [n w ɑː] Occurrences: 14,837 Examples: * ault: [ɒ ɫ t] * bots: [b ɒ t s] * opera: [ɒ p ɹ ə] * octa: [ɒ k t ə] Occurrences: 5,758 Examples: * borer: [b ɒː ɹ ə] * corr: [kʰ ɒː] * ports: [pʰ ɒː t s] * pau: [pʰ ɒː] |
Diphthongs#
aj
aw
ej
ow
ɔj
əw