German MFA dictionary v2.0.0a#
@techreport{mfa_german_mfa_dictionary_2022,
author={McAuliffe, Michael and Sonderegger, Morgan},
title={German MFA dictionary v2.0.0a},
address={\url{https://mfa-models.readthedocs.io/pronunciation dictionary/German/German MFA dictionary v2_0_0a.html}},
year={2022},
month={May},
}
G2P models Acoustic models |
Installation#
Install from the MFA command line:
mfa model download dictionary german_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 German transcripts.
This dictionary uses the MFA phone set for German, and was used in training the German 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 |
Glottal |
---|---|---|---|---|---|---|---|---|
Nasal |
Occurrences: 67,155 Examples: * kosma: [k ʰ ɔ s m a] * momo: [m ɔ m o ː] * emu: [e ː m u ː] * mukös: [m ʊ c ʰ ø ː s] Occurrences: 1,606 Examples: * azam: [a t s m ̩] * odem: [o ː d m ̩] * armem: [a ʁ m m ̩] * nahem: [n a ː m ̩] |
Occurrences: 150,364 Examples: * kirn: [c ʰ ɪ ʁ n] * binh: [b ɪ n] * maden: [m a ː d n ̩] * sagen: [z a ː ɡ n ̩] Occurrences: 68,012 Examples: * mixen: [m ɪ k s n ̩] * roten: [ʁ o ː t ʰ n ̩] * raven: [ʁ a v n ̩] * lauen: [l a w n ̩] |
Occurrences: 1,798 Examples: * singh: [z ɪ ɲ ɟ e ː h a ː] * winch: [v i ː ɲ ç] * lyng: [l ʏ ɲ ɟ e ː] * enki: [ɛ ɲ c ʰ ɪ] |
Occurrences: 31,491 Examples: * junk: [j ʊ ŋ k] * kong: [k ʰ ɔ ŋ] * lanke: [l a ŋ k ʰ ə] * ango: [a ŋ ɡ ɔ] |
||||
Stop |
Occurrences: 28,849 Examples: * kapo: [k ʰ a p ʰ ɔ] * labt: [l a ː p t] * chumb: [t ʃ ʊ m p] * pisco: [p ʰ ɪ s k ɔ] Occurrences: 60,984 Examples: * bule: [b u ː l ə] * bata: [b a t ʰ a] * bande: [b a n d ə] * bubas: [b u ː b a s] |
Occurrences: 114,741 Examples: * mitt: [m ɪ t] * gunst: [ɡ ʊ n s t] * tosen: [t ʰ o ː z n ̩] * tadel: [t ʰ a ː d l ̩] Occurrences: 53,781 Examples: * maeda: [m e ː d e ː a ː] * desto: [d ɛ s t ɔ] * pedi: [p ʰ ɛ d ɪ] * ägide: [ɛ ɟ i ː d ə] |
Occurrences: 1,183 Examples: * keks: [c ʰ e ː k s] * kür: [c ʰ y ː ɐ] * kühlt: [c ʰ y ː l t] * kilda: [c ʰ ɪ l d a ː] Occurrences: 10,500 Examples: * gavin: [ɟ ɛ v ɪ n] * gicht: [ɟ ɪ ç t] * gig: [ɟ ɪ ç] * game: [ɟ e ː m] |
Occurrences: 45,719 Examples: * brax: [b ʁ a k s] * zilog: [t s ɪ l o ː k] * drück: [d ʁ ʏ k] * lykon: [l ʏ k ʰ ɔ n] Occurrences: 56,540 Examples: * fagen: [f a ː ɡ n ̩] * sauge: [z a w ɡ ə] * gosen: [ɡ o ː z n ̩] * meges: [m ɛ ɡ ə s] |
||||
Affricate |
Occurrences: 3,855 Examples: * köpf: [c ʰ œ p f] * klapf: [k l a p f] * züpfe: [t s ʏ p f ə] * tropf: [t ʁ ɔ p f] |
Occurrences: 42,560 Examples: * cyan: [t s ʏ a ː n] * zocha: [t s ɔ x a ː] * walze: [v a l t s ə] * todds: [t ʰ o ː t s] |
Occurrences: 1,425 Examples: * chief: [t ʃ i ː f] * ruths: [ʁ ʊ t ʃ] * chaco: [t ʃ a k ʰ ɔ] * chubb: [t ʃ a p] |
|||||
Sibilant |
Occurrences: 104,057 Examples: * kals: [k ʰ a ː l s] * tast: [t ʰ a s t] * akers: [a ː k ʰ ɐ s] * maust: [m a w s t] Occurrences: 38,940 Examples: * suren: [z u ː ʁ n ̩] * käsig: [c ʰ e ː z ɪ k] * sans: [z a n s] * sinto: [z ɪ n t ʰ ɔ] |
Occurrences: 61,445 Examples: * shea: [ʃ ɛ a ː] * rotch: [ʁ o ː t ʃ] * step: [ʃ t ɛ p] * chefs: [ʃ ɛ f s] |
||||||
Fricative |
Occurrences: 72,786 Examples: * aviv: [a v i ː f] * bfz: [b e ː ɛ f t s ɛ t] * float: [f l ɔ a ː t] * gafft: [ɡ a f t] Occurrences: 41,312 Examples: * zwölf: [t s v œ l f] * raven: [ʁ a v n ̩] * boves: [b o ː v ə s] * opava: [o ː p ʰ a v a] |
Occurrences: 29,930 Examples: * äbich: [e ː b ɪ ç] * finch: [f ɪ ɲ ç] * berg: [b ɛ ɐ ç] * paich: [p ʰ a j ç] |
Occurrences: 149,619 Examples: * loral: [l ɔ ʁ a ː l] * krähe: [k ʁ e ː ə] * orda: [ɔ ʁ d a] * zider: [t s i ː d ə ʁ] |
Occurrences: 30,170 Examples: * himal: [h ɪ m a ː l] * hüst: [h y ː s t] * hem: [h ɛ m] * hinze: [h ɪ n t s ə] |
||||
Approximant |
Occurrences: 10,810 Examples: * kojen: [k ʰ o ː j n ̩] * cuyo: [k ʰ ʊ j ɔ] * seife: [z a j f ə] * seidl: [z a j t l] |
|||||||
Lateral |
Occurrences: 119,596 Examples: * klinz: [k l ɪ n t s] * säle: [z e ː l ə] * kadel: [k ʰ a ː d l ̩] * loja: [l o ː j a] Occurrences: 8,726 Examples: * disl: [d i ː z l ̩] * egel: [e ː ɡ l ̩] * nakel: [n a ː k ʰ l ̩] * hobel: [h o ː b l ̩] |
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: 0 Examples: * sepik: [z ɛ p ʰ i ː k] * ägide: [ɛ ɟ i ː d ə] * vagil: [v a ɟ i ː l] * motiv: [m ɔ t ʰ i ː f] Occurrences: 38,207 Examples: * mih: [m i ː] * venia: [v ɛ n i ː a] * egil: [ɛ ɟ i ː l] * iris: [i ː ʁ ɪ s] Occurrences: 0 Examples: * güsen: [ɟ y ː z n ̩] * grün: [ɡ ʁ y ː n] * früge: [f ʁ y ː ɡ ə] * passy: [p ʰ a s y ː] Occurrences: 9,547 Examples: * taryn: [t ʰ a ʁ y ː n] * blüm: [b l y ː m] * blüht: [b l y ː t] * rügen: [ʁ y ː ɡ n ̩] |
Occurrences: 0 Examples: * new: [ɲ u ː] * urdu: [u ː ɐ d u ː] * lut: [l u ː t] * juni: [j u ː n ɪ] Occurrences: 18,781 Examples: * flood: [f l u ː t] * nadu: [n a d u ː] * zogu: [t s o ː k ʰ u ː] * tout: [t ʰ u ː] |
|||
Occurrences: 118,412 Examples: * macy: [m a s ɪ] * dotti: [d ɔ t ʰ ɪ] * aidan: [ɛ ɪ d a ː n] * darby: [d a ʁ b ɪ] Occurrences: 12,889 Examples: * boi: [b ɔ ʏ] * brüx: [b ʁ ʏ k s] * nyro: [n ʏ ʁ ɔ] * byrne: [b ʏ ʁ n ə] |
Occurrences: 60,450 Examples: * lueg: [l ʊ e ː k] * duino: [d ʊ i ː n ɔ] * luna: [l ʊ n a ː] * pinus: [p ʰ ɪ n ʊ s] |
||||
Close-Mid |
Occurrences: 0 Examples: * gewen: [ɡ ə v e ː n] * hebet: [h e ː b e ː t] * säbel: [z e ː b l ̩] * gsell: [ɟ e ː ɛ s ɛ l] Occurrences: 39,374 Examples: * peso: [p ʰ e ː z ɔ] * cpu: [t s e ː p ʰ e ː u ː] * bka: [b e ː k ʰ a ː a ː] * bfj: [b e ː ɛ f j] Occurrences: 0 Examples: * rötet: [ʁ ø ː t ʰ ə t] * möse: [m ø ː z ə] * özlem: [ø ː t s l m ̩] * römpp: [ʁ ø ː m p] Occurrences: 5,781 Examples: * löws: [l ø ː v ə s] * söhn: [z ø ː n] * özcan: [ø ː t s k a ː n] * köniz: [c ʰ ø ː n ɪ t s] |
Occurrences: 0 Examples: * poona: [p ʰ o ː n a] * tors: [t ʰ o ː ɐ s] * ozma: [o ː t s m a] * tote: [t ʰ o ː t ʰ ə] Occurrences: 32,405 Examples: * yonus: [j o ː n ʊ s] * aoni: [a ː ɛ o ː n ɪ] * nose: [n o ː z ə] * koh: [k ʰ o ː] |
|||
Occurrences: 121,678 Examples: * boehe: [b ɔ e ː ə] * laffe: [l a f ə] * gnade: [ɡ n a ː d ə] * monde: [m o ː n d ə] |
|||||
Open-Mid |
Occurrences: 102,701 Examples: * erft: [ɛ ʁ f t] * dea: [d ɛ a ː] * mgn: [ɛ m ɟ e ː ɛ n] * riley: [ʁ ɪ j ɛ ɪ] Occurrences: 3,613 Examples: * förde: [f œ ʁ d ə] * börge: [b œ ʁ ɡ ə] * vörös: [f œ ʁ ø ː s] * corse: [c ʰ œ ʁ z ə] |
Occurrences: 61,786 Examples: * orfeo: [o ː ɔ] * duino: [d ʊ i ː n ɔ] * opole: [ɔ p ʰ o ː l ə] * vilho: [v ɪ l h ɔ] |
|||
Occurrences: 81,496 Examples: * rühr: [ʁ y ː ɐ] * leyer: [l ɛ a j ɐ] * heger: [h e ː ɡ ɐ] * mihr: [m ɐ] |
|||||
Open |
Occurrences: 112,857 Examples: * waber: [v a ː b ɐ] * salud: [z a l u ː t] * braga: [b ʁ a ɡ a] * base: [b a ː z ə] Occurrences: 46,709 Examples: * adwa: [a t v a ː] * raja: [ʁ a j a ː] * canal: [k ʰ a n a ː l] * baad: [b a ː t] |
Diphthongs#
aj
aw
ɔʏ