French MFA dictionary v3.0.0#
@techreport{mfa_french_mfa_dictionary_2024,
author={McAuliffe, Michael and Sonderegger, Morgan},
title={French MFA dictionary v3.0.0},
address={\url{https://mfa-models.readthedocs.io/pronunciation dictionary/French/French MFA dictionary v3_0_0.html}},
year={2024},
month={Mar},
}
G2P models Acoustic models |
Installation#
Install from the MFA command line:
mfa model download dictionary french_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/french/mfa/French MFA dictionary v3_0_0.dict).
Intended use#
This dictionary is intended for forced alignment of French transcripts.
This dictionary uses the MFA phone set for French, and was used in training the French 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 |
Alveolar |
Alveopalatal |
Palatal |
Velar |
Uvular |
---|---|---|---|---|---|---|---|
Nasal |
Occurrences: 21,152 Examples: * cuomo: [k ɥ ɔ m o] * malko: [m a l k o] * molas: [m ɔ l a] * röhm: [ʁ œ o m] Occurrences: 264 Examples: * mion: [mʲ ɔ̃] * mious: [mʲ u] * mien: [mʲ ɛ̃] * amiel: [a mʲ ɛ l] |
Occurrences: 18,058 Examples: * tacna: [t a k n a] * canoë: [k a n ɔ e] * niel: [n j ɛ l] * sinan: [s i n ɑ̃] |
Occurrences: 5,572 Examples: * eniwa: [ɑ̃ ɲ i w a] * sagne: [s a ɲ] * unic: [y ɲ i k] * venir: [v ə ɲ i ʁ] |
Occurrences: 673 Examples: * ying: [j i ŋ] * tông: [t o ŋ ɡ] * bing: [b i ŋ ɡ] * tring: [t ʁ i ŋ] |
|||
Stop |
Occurrences: 17,850 Examples: * tupac: [t u p a k] * poly: [p ɔ ʎ i] * parke: [p a ʁ c e] * spora: [s p ɔ ʁ a] Occurrences: 15,185 Examples: * bimbo: [b j ɛ̃ b o] * berre: [b ɛ ʁ] * buzz: [b y z] * blogs: [b l ɔ ɡ] |
Occurrences: 36,289 Examples: * batak: [b a t a k] * rotin: [ʁ o t ɛ̃] * potro: [p ɔ t ʁ o] * rôtir: [ʁ o t i ʁ] Occurrences: 19,150 Examples: * bald: [b a l d] * déçus: [d e s y] * duval: [d y v a l] * dixon: [d i z ɔ̃] |
Occurrences: 3,116 Examples: * kerk: [c ɛ ʁ k] * take: [t a c e] * kilos: [c i l ɔ s] * sakai: [s a c e] Occurrences: 969 Examples: * guus: [ɟ y] * guide: [ɟ i d] * gimié: [ɟ i mʲ e] * bougy: [b u ɟ i] |
Occurrences: 26,384 Examples: * sink: [s i ŋ k] * zac: [z a k] * haïku: [a j k u] * tics: [t i k] Occurrences: 8,609 Examples: * goûte: [ɡ u t] * aigue: [ɛ ɡ] * langa: [l ɑ̃ ɡ a] * grote: [ɡ ʁ ɔ t] |
|||
Affricate |
Occurrences: 385 Examples: * koltz: [k ɔ l ts] * seitz: [s ɛ ts] * hartz: [a ʁ ts] * motz: [m o ts] |
Occurrences: 366 Examples: * tchad: [tʃ a d] * ciot: [tʃ j o] * roth: [ʁ o tʃ] * chuck: [tʃ œ k] Occurrences: 238 Examples: * dodge: [d ɔ dʒ] * moji: [m o dʒ i] * banjo: [b ɑ̃ dʒ o] * edj: [ɛ dʒ] |
|||||
Sibilant |
Occurrences: 33,366 Examples: * croci: [k ʁ ɔ s i] * peire: [p e s ɛ ʁ] * sont: [s ɔ̃] * fixes: [f i k s] Occurrences: 9,718 Examples: * basé: [b a z e] * visas: [v i z ɑ] * muñoz: [m œ ɲ z e] * suzon: [s y z ɔ̃] |
Occurrences: 5,914 Examples: * chaux: [ʃ o] * chaut: [ʃ o] * shui: [ʃ ɥ i] * fâche: [f ɑ ʃ] Occurrences: 7,985 Examples: * mejía: [m ə ʒ i a] * jihad: [ʒ i a d] * jakob: [ʒ a k ɔ b] * ogier: [ɔ ʒ j e] |
|||||
Fricative |
Occurrences: 10,581 Examples: * refit: [ʁ ə f i] * faou: [f a u] * fusée: [f y z e] * fumet: [f y m ɛ] Occurrences: 11,456 Examples: * pskov: [p s k ɔ v] * viva: [v i v a] * avéra: [a v e ʁ a] * wetar: [v ɛ t a ʁ] |
Occurrences: 60,359 Examples: * weert: [v ɛ ʁ] * basri: [b a s ʁ i] * barra: [b a ʁ a] * verni: [v ɛ ʁ ɲ i] |
|||||
Approximant |
Occurrences: 5,009 Examples: * wigan: [w i ɡ ɑ̃] * louie: [l w i] * douar: [d w a ʁ] * voie: [v w ɑ] |
Occurrences: 14,818 Examples: * polje: [p ɔ j ʒ] * trios: [t ʁ i j o] * payer: [p e j e] * cieux: [s j ø] Occurrences: 1,898 Examples: * huan: [ɥ a n] * juéry: [ʒ ɥ e ʁ i] * druet: [d ʁ ɥ ɛ] * suais: [s ɥ ɛ] |
|||||
Lateral |
Occurrences: 27,204 Examples: * pfalz: [p e ɛ f a l z] * else: [ɛ l s] * leff: [l ɛ f] * flute: [f l y t] |
Occurrences: 6,242 Examples: * lizy: [ʎ i z i] * cali: [k a ʎ i] * salir: [s a ʎ i ʁ] * salim: [s a ʎ i m] |
Vowels#
Vowel symbols to the left of are unrounded and those to the right are rounded.
Oral Vowels#
Front |
Near-Front |
Central |
Near-Back |
Back |
|
---|---|---|---|---|---|
Close |
Occurrences: 46,665 Examples: * aoric: [a ɔ ʁ i k] * lotie: [l ɔ s i] * leoni: [l ɔ ɲ i] * ngari: [n ɡ a ʁ i] Occurrences: 12,489 Examples: * reçut: [ʁ ə s y] * tulle: [t y l] * luni: [l y ɲ i] * burne: [b y ʁ n] |
Occurrences: 7,385 Examples: * ascou: [a s k u] * ould: [u l d] * footy: [f u t i] * boue: [b u] |
|||
Close-Mid |
Occurrences: 32,704 Examples: * trié: [t ʁ i j e] * elly: [e ʎ i] * déca: [d e k a] * vérac: [v e ʁ a k] Occurrences: 1,686 Examples: * neue: [n ø] * kelme: [c e l e ɛ m ø] * jeudi: [ʒ ø d i] * romeu: [ʁ ɔ m ø] |
Occurrences: 9,655 Examples: * röhm: [ʁ œ o m] * carlo: [k a ʁ l o] * esso: [e s o] * secco: [s ə e k o] |
|||
Occurrences: 6,737 Examples: * jetée: [ʒ ə t e] * revit: [ʁ ə v i] * velay: [v ə l ɛ] * votre: [v ɔ t ʁ ə] |
|||||
Open-Mid |
Occurrences: 30,700 Examples: * weigl: [w ɛ ɡ l] * hoëne: [w ɛ n] * neil: [n ɛ j] * expos: [ɛ k s p o z] Occurrences: 2,973 Examples: * löw: [l œ] * leung: [l œ ŋ ɡ] * smuul: [s m œ œ l] * crumb: [k ʁ œ m b] |
Occurrences: 26,421 Examples: * sori: [s ɔ ʁ i] * toto: [t ɔ t o] * ora: [ɔ ʁ a] * roc'h: [ʁ ɔ] |
|||
Open |
Occurrences: 57,649 Examples: * rate: [ʁ a t] * karl: [k a ʁ l] * jemâa: [ʒ ə m ɑ a] * cima: [s i m a] |
Occurrences: 1,678 Examples: * proie: [p ʁ w ɑ] * linas: [ʎ i n ɑ] * cadre: [k ɑ d ʁ ə] * bâclé: [b ɑ k l e] |
Nasal Vowels#
Front |
Near-Front |
Central |
Near-Back |
Back |
|
---|---|---|---|---|---|
Close |
|||||
Close-Mid |
|||||
Open-Mid |
Occurrences: 7,370 Examples: * keren: [c e ʁ ɛ̃] * cazin: [k a z ɛ̃] * xingu: [k s ɛ̃ ɡ y] * armin: [a ʁ m ɛ̃] |
Occurrences: 10,708 Examples: * front: [f ʁ ɔ̃] * suwon: [s y s w ɔ̃] * monts: [m ɔ̃] * donc: [d ɔ̃ k] |
|||
Open |
Occurrences: 17,105 Examples: * landy: [l ɑ̃ d i] * medan: [m ɛ d ɑ̃] * manco: [m ɑ̃ k o] * vento: [v ɑ̃ t o] |