English MFA dictionary v2.2.1#
@techreport{mfa_english_mfa_dictionary_2023,
author={McAuliffe, Michael and Sonderegger, Morgan},
title={English MFA dictionary v2.2.1},
address={\url{https://mfa-models.readthedocs.io/pronunciation dictionary/English/English MFA dictionary v2_2_1.html}},
year={2023},
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](https://raw.githubusercontent.com/MontrealCorpusTools/mfa-models/main/dictionary/english/mfa/English MFA dictionary v2_2_1.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: 39,779 Examples: * hon: [h m ɔ n] * pomp: [pʰ ɔ m p] * limer: [l aj m ə] * scams: [s k a m z] Occurrences: 8,153 Examples: * sami: [s ɑː mʲ i] * mute: [mʲ uː t] * foamy: [f ow mʲ i] * ameba: [a mʲ iː b a] Occurrences: 3 Examples: |
Occurrences: 77,890 Examples: * haven: [ç e v ɛ n] * hench: [h ɛ n tʃ] * jeans: [dʒ iː n z] * joned: [dʒ ow n d] Occurrences: 72 Examples: |
Occurrences: 15,838 Examples: * bonny: [b ɑ ɲ i] * dini: [dʲ iː ɲ i] * handy: [h a ɲ dʲ i] * joni: [dʒ ow ɲ i] |
Occurrences: 15,459 Examples: * huang: [w ɑ ŋ] * fling: [f ʎ i ŋ ɡ] * angst: [a ŋ k s t] * ingle: [ɪ ŋ ɡ ə ɫ] |
||||
Stop |
Occurrences: 20,995 Examples: * shlep: [ʃ l ɛ p] * aspis: [a s p ɪ s] * scrip: [s c ɹ i p] * helps: [h ɛ ɫ p s] Occurrences: 25,566 Examples: * medb: [m ɛ b] * back: [b æ k] * banat: [b ə ɲ æ t] * bogie: [b ow ɟ i] |
Occurrences: 98 Examples: Occurrences: 103 Examples: |
Occurrences: 52,344 Examples: * doped: [d ow p t] * vitro: [vʲ iː t ɹ ow] * mott: [m ɑ t] * octa: [ɒ k t ə] Occurrences: 39,638 Examples: * maud: [m ɒː d] * dust: [d ɐ s t] * roid: [ɹ ɔj d] * fraud: [f ɹ ɒ d] |
Occurrences: 9,117 Examples: * mcrae: [m ə c ɹ ej] * tacky: [tʰ æ c i] * cleef: [c ʎ iː f] * skink: [s c ɪ ŋ k] Occurrences: 4,540 Examples: * ogry: [əw ɟ ɹ i] * bogey: [b ow ɟ i] * pogey: [pʰ o ɟ i] * grep: [ɟ ɹ ɛ p] |
Occurrences: 32,534 Examples: * inked: [ɪ ŋ k t] * likes: [l aj k s] * hoch: [h ɑ k] * fixes: [fʲ ɪ k s ɪ z] Occurrences: 17,059 Examples: * digha: [d aj ɡ ɑː] * hang: [h a ŋ ɡ] * glued: [ɡ l uː ɾ] * groom: [ɡ ɹ ʉː m] |
Occurrences: 435 Examples: * that: [ð a ʔ] * jet: [dʒ ɛ ʔ] * sight: [s aj ʔ] * rate: [ɹ ej ʔ] |
||
Affricate |
Occurrences: 7,118 Examples: * macho: [m ɑ tʃ ow] * chu: [tʃ u] * chalk: [tʃ ɔ k] * chevy: [tʃ ɛ vʲ i] Occurrences: 11,759 Examples: * drank: [dʒ æ ŋ k] * bejan: [bʲ iː dʒ ə n] * geo: [dʒ i ow] * joann: [dʒ əw a n] |
|||||||
Sibilant |
Occurrences: 81,177 Examples: * clits: [c ʎ ɪ t s] * afc: [ej ɛ f s iː] * bacs: [b æ k s] * syria: [s ɪ ɹ i ə] Occurrences: 33,048 Examples: * lodz: [l ɒ d z] * rinds: [ɹ aj n d z] * cise: [s aj z] * starz: [s t ɑː z] |
Occurrences: 16,175 Examples: * mesh: [m ɛ ʃ] * plush: [p l ɐ ʃ] * ashen: [æ ʃ ɪ n] * shod: [ʃ ɑ d] Occurrences: 893 Examples: * azusa: [a ʒ ʉː z ə] * uzhe: [j ʉː ʒ] * azure: [a ʒ ɒː] * niger: [ɲ iː ʒ ɛ ɹ] |
||||||
Fricative |
Occurrences: 18,520 Examples: * rfk: [ɑ ɹ ɛ f cʰ ej] * lief: [ʎ iː f] * stfan: [s t ɛ f æ n] * ruft: [ɹ ɔ f t] Occurrences: 5,062 Examples: * flick: [fʲ ʎ ɪ k] * feu: [fʲ ʉː] * fink: [fʲ i ŋ k] * filly: [fʲ ɪ ʎ i] Occurrences: 3 Examples: * fwoom: [fʷ ʉː m] Occurrences: 13,831 Examples: * valle: [v a ʎ iː] * clive: [k l aj v] * wave: [w e v] * var: [v ɑ ɹ] Occurrences: 3,112 Examples: * vela: [vʲ iː l ə] * vena: [vʲ iː n ə] * heavy: [h ɛ vʲ i] * livid: [ʎ ɪ vʲ ɪ d] Occurrences: 8 Examples: * vworp: [vʷ ɒː p] |
Occurrences: 5,857 Examples: * thay: [θ ej] * berth: [b aː θ] * maths: [m æ θ s] * oaths: [əw θ s] Occurrences: 1,368 Examples: * oaths: [əw ð z] * thy: [ð aj] * their: [ð ɛ ɹ] * thou: [ð əw] |
Occurrences: 2,158 Examples: * hinny: [ç i ɲ i] * waihi: [w aj ç iː] * gyro: [ç ɪ ɹ ow] * hick: [ç i k] |
Occurrences: 8,540 Examples: * hmm: [h ə m] * nihoa: [ɲ iː h əw ə] * whom: [h uː m] * hoyle: [h ɔj ɫ] |
||||
Approximant |
Occurrences: 9,197 Examples: * blois: [b ɫ w a] * wined: [w aj n d] * wonka: [w ɐ ŋ k ə] * whos: [w ʉː z] |
Occurrences: 69,939 Examples: * paria: [pʰ ə ɹ aj ə] * bargo: [b ɑ ɹ ɡ ow] * forma: [f ɒ ɹ m ə] * braai: [b ɹ aj] |
Occurrences: 3,280 Examples: * utica: [j ʉː tʲ ɪ k ə] * yee: [j i] * yer: [j ɝ] * abaya: [ə b ej j ə] |
|||||
Tap |
Occurrences: 478 Examples: * read: [ɹ ɛ ɾ] * code: [kʰ əw ɾ] * honor: [ɒ ɾ̃ ə] * lord: [l ɒː ɾ] Occurrences: 70 Examples: * ready: [ɹ ɛ ɾʲ i] * idea: [aj ɾʲ j ə] * added: [a ɾʲ ɪ ɾ] * body: [b ɒ ɾʲ i] Occurrences: 67 Examples: |
|||||||
Lateral |
Occurrences: 29,417 Examples: * slur: [s l ɝ] * plays: [p l e z] * claus: [k l ɒː s] * malar: [m ej l ə] Occurrences: 24,653 Examples: * ideal: [aj dʲ iː a ɫ] * tile: [tʰ aj ɫ] * jell: [dʒ ɛ ɫ] * mohel: [m əw ɛ ɫ] Occurrences: 58 Examples: |
Occurrences: 20,831 Examples: * listi: [ʎ i s tʲ i] * oglio: [əw ʎ əw] * paley: [pʰ ej ʎ i] * linc: [ʎ ɪ ŋ k] |
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: 63,712 Examples: * petri: [pʰ ɛ tʲ ɹ i] * wiley: [w aj ʎ i] * cali: [cʰ æ ʎ i] * beef: [bʲ i f] Occurrences: 17,090 Examples: * slc: [ɛ s ɛ ɫ s iː] * ipcc: [aj pʰ iː s iː s iː] * chico: [ʃ iː k ow] * gci: [dʒ iː s iː aj] |
Occurrences: 2,829 Examples: * nehru: [ɲ ej ɹ ʉ] * sotho: [s ʉ tʰ ʉ] * fluke: [f l ʉ k] * mutsu: [m ʊ t s ʉ] Occurrences: 8,244 Examples: * educe: [ɪ dʲ ʉː s] * loom: [l ʉː m] * supra: [s ʉː p ɹ ə] * ploom: [p l ʉː m] |
Occurrences: 4,034 Examples: * lujvo: [l u ʃ v o] * apus: [e p u s] * caput: [kʰ a p u t] * sinus: [s aj n u s] Occurrences: 4,665 Examples: * taboo: [tʰ a b uː] * neuk: [ɲ uː k] * tuam: [tʰ uː a m] * vertu: [v aː tʰ uː] |
||
Occurrences: 72,843 Examples: * blink: [b ʎ ɪ ŋ k] * akin: [ɐ c ɪ n] * silt: [s ɪ ɫ t] * grim: [ɟ ɹ ɪ m] |
Occurrences: 8,061 Examples: * uttar: [ʊ t ɚ] * tunde: [tʰ ʊ n d e] * lure: [ʎ ʊ ə] * mure: [m ʊ ɹ] |
||||
Close-Mid |
Occurrences: 7,660 Examples: * rave: [ɹ e v] * acog: [e kʰ a ɡ] * phase: [f e z] * meiji: [m e dʒ i] Occurrences: 17,268 Examples: * bae: [b ej] * naomi: [ɲ ej ə mʲ i] * copay: [kʰ ow pʰ ej] * mayor: [m ej ɹ] |
Occurrences: 7,002 Examples: * soc: [s o ʃ] * yobbo: [j ɔ b o] * owens: [o w ɛ n z] * bone: [b o n] Occurrences: 10,089 Examples: * boden: [b ow d ə n] * sonia: [s ow ɲ i ə] * throw: [θ ɹ ow] * tote: [tʰ ow t] |
|||
Occurrences: 86,311 Examples: * ryan: [ɹ aj ə n] * berat: [b ə ɹ a t] * ongar: [ɒ ŋ ɡ ə] * haifa: [ç ej f ə] Occurrences: 9,815 Examples: * liter: [ʎ iː t ɚ] * tuner: [tʰ ʉː n ɚ] * your: [j ɚ] * enter: [ɛ n t ɚ] |
|||||
Open-Mid |
Occurrences: 46,399 Examples: * fret: [f ɹ ɛ t] * tens: [tʰ ɛ n z] * yet's: [j ɛ t s] * nello: [n ɛ l ow] Occurrences: 1,565 Examples: * bear: [b ɛː] * khmer: [kʰ ə m ɛː] * aires: [ɛː z] * aire: [ɛː] |
Occurrences: 308 Examples: * murr: [m ɜ] * furor: [fʲ ɜ ɹ ə] * circe: [s ɜ s i] * turku: [tʰ ɜ k ʉː] Occurrences: 3,191 Examples: * pasir: [pʰ ə s ɜː] * yrneh: [ɜː ɲ i] * wurtz: [w ɜː t s] * wurst: [v ɜː s t] Occurrences: 3,490 Examples: * viera: [vʲ iː ɝ ə] * vert: [v ɝ t] * fern: [f ɝ n] * rural: [ɹ ɝ ə ɫ] |
Occurrences: 23,458 Examples: * tubby: [tʰ ɔ bʲ i] * maud: [m ɔ d] * pork: [pʰ ɔ k] * agora: [a ɡ ɔ ɹ a] |
||
Occurrences: 15,904 Examples: * slams: [s l æ m z] * graf: [ɡ ɹ æ f] * patty: [pʰ æ tʲ i] * acids: [æ s ɪ d z] |
Occurrences: 10,496 Examples: * sudds: [s ɐ d z] * rutch: [ɹ ɐ tʃ] * zuppa: [z ɐ p ə] * brush: [b ɹ ɐ ʃ] |
||||
Open |
Occurrences: 47,485 Examples: * arc: [a k] * caron: [kʰ a ə n] * mega: [m ɛ ɡ a] * fanon: [f a n u n] Occurrences: 2,143 Examples: * yrneh: [aː ɲ i] * overt: [o v aː t] * earl: [aː ɫ] * adieu: [a dʲ aː] |
Occurrences: 11,027 Examples: * tonia: [tʰ ɑ ɲ i ə] * zohar: [z ow h ɑ ɹ] * dobbs: [d ɑ b z] * jot: [dʒ ɑ t] Occurrences: 7,178 Examples: * short: [ʃ ɑː t] * horse: [h ɑː s] * ala: [ɑː l ə] * anput: [ɑː n pʰ ʊ t] Occurrences: 14,872 Examples: * goth: [ɡ ɒ θ] * ogle: [ɒ ɡ ə ɫ] * frock: [f ɹ ɒ k] * cores: [kʰ ɒ ɹ z] Occurrences: 5,776 Examples: * yaws: [j ɒː z] * audi: [ɒː dʲ i] * utahn: [j ʉː tʰ ɒː n] * porto: [pʰ ɒː tʰ əw] |
Diphthongs#
aj
aw
ej
ow
ɔj
əw