Swedish MFA dictionary v2.0.0#
@techreport{mfa_swedish_mfa_dictionary_2022,
author={McAuliffe, Michael and Sonderegger, Morgan},
title={Swedish MFA dictionary v2.0.0},
address={\url{https://mfa-models.readthedocs.io/pronunciation dictionary/Swedish/Swedish MFA dictionary v2_0_0.html}},
year={2022},
month={Mar},
}
G2P models Acoustic models |
Installation#
Install from the MFA command line:
mfa model download dictionary swedish_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 Swedish transcripts.
This dictionary uses the MFA phone set for Swedish, and was used in training the Swedish 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 |
Retroflex |
Palatal |
Velar |
Glottal |
---|---|---|---|---|---|---|---|---|
Nasal |
Occurrences: 14,317 Examples: * emot: [eː m uː t̪] * gusum: [ɡ ɵ s̪ ɵ m] * tomas: [t̪ʰ ɔ˥˧ m a˥˩ s̪] * modem: [m uː˧˩ d̪ eː m] |
Occurrences: 39,682 Examples: * anden: [a n̪ d̪ ɛ n̪] * tina: [t̪ʰ iː˧˩ n̪ a˥˩] * bion: [b ɪ ʊ n̪] * and: [a n̪ d̪] |
Occurrences: 2,292 Examples: * erna: [ɛ ɳ a] * torn: [t̪ʰ uː ɳ] * ahern: [ɑː ɛ ɳ] * hearn: [h ɛ a ɳ] |
Occurrences: 7,498 Examples: * angår: [a ŋ ɡ oː r] * zink: [s̪ ɪ ŋ k] * young: [yː uː ɵ ŋ] * säng: [s̪ ɛ ŋ] |
||||
Stop |
Occurrences: 7,258 Examples: * spä: [s̪ p ɛ] * oppe: [ɔ p ɛ˥˩] * köpta: [ɕ øː p t̪ a] * spred: [s̪ p r eː d̪] Occurrences: 9,730 Examples: * bruce: [b r ʉː k eː˧˩] * bli: [b l iː] * bion: [b ɪ ʊ n̪] * bredd: [b r eː˥˩ d̪] |
Occurrences: 31,064 Examples: * etisk: [eː t̪ ɪ s̪ k] * dalat: [d̪ ɑː˧˩ l a˥˩ t̪] * skit: [ɧ iː t̪] * stick: [s̪ t̪ ɪ˧˩ k] Occurrences: 19,536 Examples: * anden: [a n̪ d̪ ɛ n̪] * dalat: [d̪ ɑː˧˩ l a˥˩ t̪] * adele: [ɑː d̪ ɛ l ɛ] * slöjd: [s̪ l œ j d̪] |
Occurrences: 1,325 Examples: * borta: [b ɔ˧˩ ʈ a] * körts: [ɕ øː ʈ s̪] * snart: [s̪ n̪ ɑː ʈ] * start: [s̪ t̪ ɑː ʈ] Occurrences: 1,760 Examples: * tärd: [t̪ʰ ɛː ɖ] * förde: [f œ ɖ ɛ] * torde: [t̪ʰ uː ɖ ɛ] * tärde: [t̪ʰ ɛː ɖ ɛ] |
Occurrences: 18,094 Examples: * kväll: [k ʋ ɛ l] * bruce: [b r ʉː k eː˧˩] * etisk: [eː t̪ ɪ s̪ k] * stick: [s̪ t̪ ɪ˧˩ k] Occurrences: 10,087 Examples: * wååg: [ʋ oː˧˩ oː ɡ] * angår: [a ŋ ɡ oː r] * gusum: [ɡ ɵ s̪ ɵ m] * kugge: [kʰ ɵ ɡ ɛ˥˩] |
||||
Sibilant |
Occurrences: 38,194 Examples: * oss: [ɔ s̪] * etisk: [eː t̪ ɪ s̪ k] * stick: [s̪ t̪ ɪ˧˩ k] * spä: [s̪ p ɛ] |
Occurrences: 2,359 Examples: * bars: [b a ʂ] * norsk: [n̪ ɔ˧˩ ʂ k] * mars: [m a ʂ] * borch: [b ɔ ʂ ɕ] |
Occurrences: 1,376 Examples: * köpta: [ɕ øː p t̪ a] * kille: [ɕ ɪ˧˩ l ɛ˥˩] * king: [ɕ ɪ ŋ] * köld: [ɕ øː l d̪] |
|||||
Fricative |
Occurrences: 10,927 Examples: * flyt: [f l yː˧˩ t̪] * kaffe: [kʰ a˧˩ f ɛ˥˩] * filar: [f ɪ l a r] * fyran: [f ʏ r ɑː n̪] |
Occurrences: 6,227 Examples: * kohls: [kʰ ʊ h l s̪] * höör: [h œ øː r] * hals: [h a˧˩ l s̪] * perth: [pʰ ɛ r t̪ h] |
||||||
Approximant |
Occurrences: 11,757 Examples: * kväll: [k ʋ ɛ l] * wååg: [ʋ oː˧˩ oː ɡ] * våras: [ʋ oː˧˩ r a˥˩ s̪] * oviss: [uː˧˩ ʋ ɪ s̪] |
Occurrences: 5,785 Examples: * djur: [j ʉː r] * slöjd: [s̪ l œ j d̪] * göte: [j œ t̪ ɛ] * sanoj: [s̪ a n̪ ʊ j] |
||||||
Trill |
Occurrences: 43,405 Examples: * djur: [j ʉː r] * bruce: [b r ʉː k eː˧˩] * tar: [t̪ʰ ɑː r] * våras: [ʋ oː˧˩ r a˥˩ s̪] |
|||||||
Lateral |
Occurrences: 25,226 Examples: * kväll: [k ʋ ɛ l] * bli: [b l iː] * dalat: [d̪ ɑː˧˩ l a˥˩ t̪] * slöts: [s̪ l øː t̪ s̪] |
|||||||
Lateral Tap |
Occurrences: 407 Examples: * sorl: [s̪ oː ɭ] * carl: [kʰ ɑː ɭ] * årlig: [oː˧˩ ɭ ɪ˥˩ ɡ] * arla: [ɑː ɭ a] |
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: 5,806 Examples: * bli: [b l iː] * skit: [ɧ iː t̪] * deri: [d̪ ɛ r iː] * bil: [b iː l] Occurrences: 1,350 Examples: * ray: [r ɑː˥˩ yː] * asyl: [a s̪ yː l] * getty: [j ɛ t̪ʰ yː] * tyra: [t̪ʰ yː r a] |
Occurrences: 2,985 Examples: * djura: [j ʉː˥˩ r a˥˩] * rus: [r ʉː s̪] * odjur: [uː˧˩ j ʉː˥˩ r] * huka: [h ʉː˥˩ k a] |
Occurrences: 4,277 Examples: * zoli: [s̪ uː l ɪ] * emot: [eː m uː t̪] * tom: [t̪ʰ uː m] * valbo: [ʋ a l b uː˥˩] |
||
Occurrences: 18,354 Examples: * bois: [b uː˥˩ ɪ˩ s̪] * etisk: [eː t̪ ɪ s̪ k] * strix: [s̪ t̪ r ɪ k s̪] * sveis: [s̪ ʋ ɛ˧˩ ɪ˩ s̪] Occurrences: 2,832 Examples: * fyran: [f ʏ r ɑː n̪] * byst: [b ʏ s̪ t̪] * lyfta: [l ʏ˧˩ f t̪ʰ a] * hyste: [h ʏ s̪ t̪ ɛ] |
Occurrences: 6,402 Examples: * disko: [d̪ ɪ s̪ k ʊ˩] * älgko: [ɛ l j k ʊ˩] * doser: [d̪ ʊ s̪ ɛ r] * cover: [k ʊ ʋ ɛ r] |
||||
Close-Mid |
Occurrences: 11,384 Examples: * emot: [eː m uː t̪] * avled: [ɑː˧˩ ʋ l eː˥˩ d̪] * svea: [s̪ ʋ eː˧˩ a˥˩] * bredd: [b r eː˥˩ d̪] Occurrences: 4,643 Examples: * strös: [s̪ t̪ r øː˥˩ s̪] * utför: [ʉː˧˩ t̪ f øː˥˩ r] * slöts: [s̪ l øː t̪ s̪] * avgör: [ɑː˧˩ ʋ j øː˥˩ r] |
Occurrences: 7,317 Examples: * tukt: [t̪ʰ ɵ k t̪] * gusum: [ɡ ɵ s̪ ɵ m] * kugge: [kʰ ɵ ɡ ɛ˥˩] * louie: [l uː ɵ iː˧˩ ɛ˥˩] |
Occurrences: 4,223 Examples: * varpå: [ʋ a r p oː˥˩] * låt: [l oː t̪] * uppå: [ɵ˥˧ p oː˩] * tom: [t̪ʰ oː m] |
||
Open-Mid |
Occurrences: 47,840 Examples: * efta: [ɛ˧˩ f t̪ a˥˩] * svepa: [s̪ ʋ ɛ˥˧ p a] * kugge: [kʰ ɵ ɡ ɛ˥˩] * träet: [t̪ r ɛ˧˩ ɛ t̪] Occurrences: 2,659 Examples: * affär: [a f ɛː˥˧ r] * ärad: [ɛː r a d̪] * klä: [k l ɛː] * näten: [n̪ ɛː t̪ ɛ˩ n̪] Occurrences: 2,880 Examples: * höör: [h œ øː r] * rört: [r œ˥˩ ʈ] * höljt: [h œ l j t̪] * slöjd: [s̪ l œ j d̪] |
Occurrences: 10,160 Examples: * pholm: [pʰ h ɔ˥˩ l m] * ronny: [r ɔ n̪ ʏ˩] * lotte: [l ɔ˧˩ t̪ ɛ˥˩] * oppe: [ɔ p ɛ˥˩] |
|||
Open |
Occurrences: 33,850 Examples: * anden: [a n̪ d̪ ɛ n̪] * angår: [a ŋ ɡ oː r] * and: [a n̪ d̪] * filar: [f ɪ l a r] |
Occurrences: 12,465 Examples: * adele: [ɑː d̪ ɛ l ɛ] * bada: [b ɑː˧˩ d̪ a˥˩] * gap: [j ɑː˥˧ p] * laval: [l ɑː˧˩ ʋ ɑː l] |
Tones#
˥˧
˥˩
˧˩
˩