English (US) MFA dictionary v2.2.1#
@techreport{mfa_english_us_mfa_dictionary_2023,
author={McAuliffe, Michael and Sonderegger, Morgan},
title={English (US) MFA dictionary v2.2.1},
address={\url{https://mfa-models.readthedocs.io/pronunciation dictionary/English/English (US) MFA dictionary v2_2_1.html}},
year={2023},
month={May},
}
G2P models |
Installation#
Install from the MFA command line:
mfa model download dictionary english_us_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 (US) 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: 17,485 Examples: * mayan: [m aj ə n] * almah: [æ ɫ m ə] * moult: [m ow ɫ t] * amite: [ej m ɛ t] Occurrences: 3,285 Examples: * agami: [æ ɡ ə mʲ i] * jemmy: [dʒ ɛ mʲ i] * memic: [mʲ i mʲ ɪ k] * mists: [mʲ ɪ s t s] Occurrences: 25 Examples: |
Occurrences: 9 Examples: |
Occurrences: 34,399 Examples: * anton: [æ n t ɑ n] * noil: [n ɔj ɫ] * kuban: [cʰ ʉː b ɑː n] * ne'er: [n ɛ ɹ] Occurrences: 236 Examples: |
Occurrences: 7,153 Examples: * jinny: [dʒ ɪ ɲ i] * enact: [ɪ ɲ æ k t] * nabbs: [ɲ æ b z] * nico: [ɲ i k ow] |
Occurrences: 7,367 Examples: * mongo: [m ɑː ŋ ɡ ow] * brung: [b ɹ ɐ ŋ] * using: [j ʉː z ɪ ŋ] * inc: [ɪ ŋ k] |
|||
Stop |
Occurrences: 8,790 Examples: * pluto: [p l ʉː t ow] * prove: [p ɹ ʉː v] * plead: [p l ɛ d] * sleep: [s ʎ i p] Occurrences: 11,342 Examples: * newbz: [ɲ ʉː b z] * soba: [s ow b ə] * abort: [ə b ɒ ɹ t] * both: [b ow θ] |
Occurrences: 32 Examples: Occurrences: 55 Examples: |
Occurrences: 23,004 Examples: * vital: [v aj t ə ɫ] * soft: [s ɒ f t] * yurt: [j ʊ ɹ t] * shoot: [ʃ ʉː t] Occurrences: 17,316 Examples: * doane: [d ow n] * dazed: [d ej z d] * rowed: [ɹ ow d] * poled: [pʰ ow ɫ d] |
Occurrences: 4,402 Examples: * lunky: [l ɐ ɲ c i] * clare: [c ʎ ɛ ɹ] * cleft: [c ʎ ɛ f t] * krebs: [c ɹ ɛ b z] Occurrences: 2,701 Examples: * gill: [ɟ ɪ ɫ] * guile: [ɟ aj ɫ] * glens: [ɟ ʎ ɛ n z] * gills: [ɟ ɪ ɫ z] |
Occurrences: 13,512 Examples: * cried: [k ɹ aj d] * excel: [ɪ k s ɛ ɫ] * pac: [pʰ æ k] * scuff: [s k ɐ f] Occurrences: 5,216 Examples: * yaga: [j ɑː ɡ ə] * agate: [ə ɡ æ t] * glory: [ɡ l ɒː ɹ i] * sigg: [s ɪ ɡ] |
Occurrences: 207 Examples: * ought: [ɒ ʔ] * rat: [ɹ æ ʔ] * often: [ɑː f ʔ ə n] * quote: [kʷ ow ʔ] |
||
Affricate |
Occurrences: 3,325 Examples: * chase: [tʃ ej s] * gulch: [ɡ ɐ ɫ tʃ] * try: [tʃ aj] * chaps: [tʃ æ p s] Occurrences: 4,970 Examples: * julep: [dʒ ʉː l ɛ p] * jawn: [dʒ ɒː n] * mage: [m ej dʒ] * jails: [dʒ ej ɫ z] |
|||||||
Sibilant |
Occurrences: 36,484 Examples: * spore: [s p ɒ ɹ] * parsi: [pʰ ɑ ɹ s i] * texed: [tʰ ɛ k s t] * skint: [s c ɪ n t] Occurrences: 15,814 Examples: * chaz: [tʃ æ z] * johns: [dʒ ɑ n z] * hymns: [ç ɪ m z] * zizz: [z ɪ z] |
Occurrences: 6,861 Examples: * shing: [ʃ ɪ ŋ] * shag: [ʃ æ ɡ] * smush: [s m ʊ ʃ] * shity: [ʃ ɪ tʲ i] Occurrences: 559 Examples: * usual: [j ʉː ʒ ɫ̩] * dejah: [d ej ʒ ə] * hajj: [h ɑ ʒ] * jete: [ʒ ɛ tʰ ej] |
||||||
Fricative |
Occurrences: 8,215 Examples: * flux: [f l ɐ k s] * leafs: [ʎ iː f s] * photo: [f ow t ow] * force: [f ɒ ɹ s] Occurrences: 2,166 Examples: * fiend: [fʲ iː n d] * flisk: [fʲ ʎ ɪ s k] * finch: [fʲ ɪ n tʃ] * fjord: [fʲ ɒ ɹ d] Occurrences: 5,607 Examples: * hev: [h ɛ v] * ivan: [i v ɑː n] * diva: [dʲ iː v ə] * lnav: [ɛ l ɲ ej v] Occurrences: 1,728 Examples: * vim: [vʲ ɪ m] * dvd: [dʲ iː vʲ iː dʲ iː] * dvds: [dʲ iː vʲ iː dʲ iː z] * vide: [vʲ iː d ej] |
Occurrences: 2,527 Examples: * thor: [θ ɒ ɹ] * thong: [θ ɑ ŋ] * kathy: [cʰ æ θ i] * death: [d ɛ θ] Occurrences: 637 Examples: * meath: [mʲ iː ð] * thems: [ð ɛ m z] * rethe: [ɹ iː ð] * oaths: [ow ð z] |
Occurrences: 976 Examples: * hail: [ç ej ɫ] * hazen: [ç ej z ə n] * hist: [ç ɪ s t] * here: [ç ɪ ɹ] |
Occurrences: 3,717 Examples: * hexad: [h ɛ k s æ d] * hatta: [h æ t ə] * hoofs: [h ʉ f s] * honey: [h ɐ ɲ i] |
||||
Approximant |
Occurrences: 4,440 Examples: * dwell: [d w ɛ ɫ] * luan: [ʎ ʉ w ɑ n] * wabi: [w ɑː bʲ i] * wirra: [w ɪ ɹ ə] |
Occurrences: 35,258 Examples: * ruff: [ɹ ɐ f] * orin: [ɒ ɹ ɪ n] * fours: [f ɒː ɹ z] * yarl: [j ɑ ɹ ɫ] |
Occurrences: 1,149 Examples: * yes: [j ɛ s] * ewen: [j ʉː ə n] * ubac: [j ʉ b æ k] * yam: [j æ m] |
|||||
Tap |
Occurrences: 139 Examples: * god: [ɡ ɑ ɾ] * gotta: [ɡ ɑ ɾ ə] * paid: [pʰ ej ɾ] * items: [aj ɾ ə m z] Occurrences: 117 Examples: * city: [s ɪ ɾʲ i] * ready: [ɹ ɛ ɾʲ i] * dated: [d ej ɾʲ ɪ d] * buddy: [b ɐ ɾʲ i] Occurrences: 36 Examples: |
|||||||
Lateral |
Occurrences: 12,204 Examples: * blush: [b l ɐ ʃ] * gloat: [ɡ l ow t] * late: [l ej t] * gusla: [ɡ ʊ s l ə] Occurrences: 11,157 Examples: * cool: [kʰ ʉː ɫ] * skulk: [s k ɐ ɫ k] * hulls: [h ɐ ɫ z] * afoul: [ə f aw ɫ] Occurrences: 110 Examples: |
Occurrences: 9,753 Examples: * sully: [s ɐ ʎ i] * glad: [ɟ ʎ æ d] * lyra: [ʎ ɪ ɹ ə] * shyly: [ʃ aj 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: 16,574 Examples: * cosey: [kʰ ow z i] * ninny: [ɲ ɪ ɲ i] * rocky: [ɹ ɑ c i] * rashy: [ɹ æ ʃ i] Occurrences: 6,400 Examples: * wfoe: [w ʊ fʲ iː] * dll: [dʲ iː ɛ ɫ ɛ ɫ] * bixie: [bʲ iː ʃ i ej] * three: [θ ɹ iː] |
Occurrences: 2,271 Examples: * sault: [s ʉ] * few: [fʲ ʉ] * jewry: [dʒ ʉ ə ɹ i] * toots: [tʰ ʉ t s] Occurrences: 4,636 Examples: * lubes: [ʎ ʉː b z] * usest: [j ʉː z ə s t] * pupa: [pʲ ʉː p ə] * whoes: [ç ʉː z] |
|||
Occurrences: 44,669 Examples: * boid: [b ow ɪ d] * optic: [ɑ p tʲ ɪ k] * limbs: [ʎ ɪ m z] * titty: [tʰ ɪ tʲ i] |
Occurrences: 2,529 Examples: * broom: [b ɹ ʊ m] * pusta: [pʰ ʊ s t ə] * woman: [w ʊ m ə n] * put: [pʰ ʊ ʔ] |
||||
Close-Mid |
Occurrences: 11,327 Examples: * slate: [s l ej t] * phage: [f ej dʒ] * plays: [p l ej z] * eish: [ej ʃ] |
Occurrences: 10,146 Examples: * doses: [d ow s i z] * doe: [d ow] * yeo: [j ow] * polo: [pʰ ow l ow] |
|||
Occurrences: 47,030 Examples: * maka: [m ɑ k ə] * poems: [pʰ ow ə m z] * uncle: [ɐ ŋ k ə ɫ] * sfnal: [ɛ s ɛ f n ə ɫ] Occurrences: 9,853 Examples: * avers: [ej v ɚ z] * major: [m ej dʒ ɚ] * loser: [ʎ ʉ z ɚ] * tenor: [tʰ ɛ n ɚ] |
|||||
Open-Mid |
Occurrences: 18,865 Examples: * hexes: [h ɛ k s ɪ z] * crepe: [c ɹ ɛ p] * deco: [d ɛ k ow] * plebs: [p l ɛ b z] |
Occurrences: 3,497 Examples: * serb: [s ɝ b] * ashur: [æ ʃ ɝ] * lured: [l ɝ d] * ferko: [f ɝ k ow] |
|||
Occurrences: 15,969 Examples: * alga: [æ ɫ ɡ ə] * zaman: [z æ m ə n] * swang: [s w æ ŋ] * babel: [b æ b ə ɫ] |
Occurrences: 7,424 Examples: * wyd: [d ɐ bʲ ʉː w aj dʲ iː] * clunk: [k l ɐ ŋ k] * null: [n ɐ ɫ] * junky: [dʒ ɐ ɲ c i] |
||||
Open |
Occurrences: 0 Examples: Occurrences: 5 Examples: * might: [m aː] * like: [l aː k] * likes: [l aː k s] * night: [n aː t] |
Occurrences: 10,648 Examples: * copse: [kʰ ɑ p s] * maja: [m ɑ j ə] * josh: [dʒ ɑ ʃ] * dion: [dʲ iː ɑ n] Occurrences: 2,436 Examples: * bari: [b ɑː ɹ i] * blois: [b l w ɑː] * nauru: [n ɑː ʉː ɹ ʉː] * basra: [b ɑː ɹ ə] Occurrences: 5,324 Examples: * osm: [ɒ s ə m] * boggs: [b ɒ ɡ z] * warps: [w ɒ ɹ p s] * tawn: [tʰ ɒ n] Occurrences: 1,559 Examples: * water: [w ɒː t ɚ] * gaul: [ɡ ɒː ɫ] * soyuz: [s ɒː j ʉː z] * haut: [h ɒː t] |
Diphthongs#
aj
aw
ej
ow
ɔj