Croatian MFA dictionary v2.0.0#
@techreport{mfa_croatian_mfa_dictionary_2022,
author={McAuliffe, Michael and Sonderegger, Morgan},
title={Croatian MFA dictionary v2.0.0},
address={\url{https://mfa-models.readthedocs.io/pronunciation dictionary/Croatian/Croatian 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 croatian_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 Serbo-Croatian transcripts.
This dictionary uses the MFA phone set for Croatian, and was used in training the Croatian 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 |
---|---|---|---|---|---|---|---|
Nasal |
Occurrences: 9,333 Examples: * magma: [m a˨˦ ɡ m a] * jamči: [j a˦˨ m tʂ i] * tulum: [t̪ u˨˦ l u m] * burma: [b uː˦˨ r m a] |
Occurrences: 20,674 Examples: * ladan: [l aː˨˦ d̪ a n] * venac: [v eː˨˦ n a t̪s̪] * novac: [n o˨˦ v a t̪s̪] * jean: [j e˨˦ aː n] |
Occurrences: 2,371 Examples: * manja: [m a ɲ a] * njeni: [ɲ e˨˦ n iː] * njem: [ɲ e˦˨ m] * munja: [m uː˨˦ ɲ a] |
||||
Stop |
Occurrences: 10,522 Examples: * priča: [p r iː˦˨ tʂ a] * petar: [p e˨˦ t̪ a r] * lopar: [l o˨˦ p aː r] * papom: [p a˦˨ p oː m] Occurrences: 5,318 Examples: * club: [t̪s̪ l u b] * baba: [b a˦˨ b a] * gabor: [ɡ aː˦˨ b o r] * burma: [b uː˦˨ r m a] |
Occurrences: 20,370 Examples: * petar: [p e˨˦ t̪ a r] * orati: [o˨˦ r a t̪ i] * kata: [k a t̪ a] * kuta: [k uː˨˦ t̪ a] Occurrences: 8,678 Examples: * ladan: [l aː˨˦ d̪ a n] * osuda: [o˦˨ s̪ u d̪ a] * drže: [d̪ r̩˨˦ ʐ eː] * duga: [d̪ uː˨˦ ɡ a] |
Occurrences: 12,670 Examples: * šaka: [ʂ a˦˨ k a] * kata: [k a t̪ a] * kenzo: [k e˦˨ n z̪ o] * kuta: [k uː˨˦ t̪ a] Occurrences: 4,907 Examples: * magma: [m a˨˦ ɡ m a] * grana: [ɡ r aː˨˦ n a] * duga: [d̪ uː˨˦ ɡ a] * gabor: [ɡ aː˦˨ b o r] |
||||
Affricate |
Occurrences: 4,150 Examples: * carlo: [t̪s̪ a˦˨ r l o] * venac: [v eː˨˦ n a t̪s̪] * novac: [n o˨˦ v a t̪s̪] * club: [t̪s̪ l u b] |
Occurrences: 1,828 Examples: * božić: [b o˨˦ ʐ i tɕ] * nosić: [n o˨˦ s̪ i tɕ] * ćuran: [tɕ u˨˦ r aː n] * ćup: [tɕ u˦˨ p] Occurrences: 687 Examples: * vjeđa: [v j e˦˨ dʑ a] * pređa: [p r e˦˨ dʑ a] * žeđ: [ʐ eː˦˨ dʑ] * đavao: [dʑ a˦˨ v a o] |
|||||
Sibilant |
Occurrences: 14,671 Examples: * osuda: [o˦˨ s̪ u d̪ a] * slici: [s̪ l i˨˦ t̪s̪ i] * elzas: [e˨˦ l z̪ a s̪] * sferi: [s̪ f e˨˦ r i] Occurrences: 6,062 Examples: * kenzo: [k e˦˨ n z̪ o] * elzas: [e˨˦ l z̪ a s̪] * uzdah: [u˦˨ z̪ d̪ aː x] * čizma: [tʂ i˦˨ z̪ m a] |
Occurrences: 3,471 Examples: * šaka: [ʂ a˦˨ k a] * šanac: [ʂ a˦˨ n a t̪s̪] * štab: [ʂ t̪ aː˦˨ b] * lemeš: [l e˨˦ m e ʂ] Occurrences: 1,782 Examples: * drže: [d̪ r̩˨˦ ʐ eː] * užoj: [u˨˦ ʐ o j] * božić: [b o˨˦ ʐ i tɕ] * ženik: [ʐ e˨˦ n iː k] |
Occurrences: 395 Examples: * sjera: [ɕ e˦˨ r a] * sjati: [ɕ a˦˨ t̪ i] * gošća: [ɡ o˨˦ ɕ tɕ a] * sijed: [ɕ eː˦˨ d̪] Occurrences: 55 Examples: * dežđ: [d̪ e ʑ dʑ] * zjena: [ʑ e˦˨ n a] * kozji: [k o˦˨ ʑ iː] |
||||
Fricative |
Occurrences: 1,314 Examples: * sferi: [s̪ f e˨˦ r i] * fokus: [f oː˨˦ k u s̪] * kifla: [k iː˨˦ f l a] * kofer: [k o˨˦ f e r] Occurrences: 11,740 Examples: * venac: [v eː˨˦ n a t̪s̪] * novac: [n o˨˦ v a t̪s̪] * vlado: [v l aː˨˦ d̪ o] * vjeđa: [v j e˦˨ dʑ a] |
||||||
Approximant |
Occurrences: 8,108 Examples: * jamči: [j a˦˨ m tʂ i] * jean: [j e˨˦ aː n] * užoj: [u˨˦ ʐ o j] * vjeđa: [v j e˦˨ dʑ a] |
||||||
Trill |
Occurrences: 19,425 Examples: * priča: [p r iː˦˨ tʂ a] * carlo: [t̪s̪ a˦˨ r l o] * petar: [p e˨˦ t̪ a r] * orati: [o˨˦ r a t̪ i] Occurrences: 32 Examples: * ubrzo: [u˦˨ b rː z̪ o] Occurrences: 322 Examples: * prvih: [p r̩˨˦ v i x] * mrkov: [m r̩˨˦ k o v] * prka: [p r̩˨˦ k a] * drže: [d̪ r̩˨˦ ʐ eː] Occurrences: 169 Examples: * brko: [b r̩ː˨˦ k o] * tvrd: [t̪ v r̩ː˦˨ d̪] * krpež: [k r̩ː˨˦ p e ʐ] * vrsta: [v r̩ː˨˦ s̪ t̪ a] |
||||||
Lateral |
Occurrences: 10,376 Examples: * ladan: [l aː˨˦ d̪ a n] * carlo: [t̪s̪ a˦˨ r l o] * club: [t̪s̪ l u b] * slici: [s̪ l i˨˦ t̪s̪ i] |
Occurrences: 2,460 Examples: * ključ: [k ʎ uː˦˨ tʂ] * ljude: [ʎ u˨˦ d̪ e] * polje: [p o˦˨ ʎ e] * ljut: [ʎ uː˦˨ t̪] |
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: 22,690 Examples: * jamči: [j a˦˨ m tʂ i] * orati: [o˨˦ r a t̪ i] * slici: [s̪ l i˨˦ t̪s̪ i] * sferi: [s̪ f e˨˦ r i] Occurrences: 7,877 Examples: * marki: [m a˦˨ r k iː] * kifla: [k iː˨˦ f l a] * irski: [iː˦˨ r s̪ k iː] * zbiva: [z̪ b iː˨˦ v a] |
Occurrences: 6,235 Examples: * osuda: [o˦˨ s̪ u d̪ a] * club: [t̪s̪ l u b] * tulum: [t̪ u˨˦ l u m] * vodu: [v o d̪ u] Occurrences: 990 Examples: * ključ: [k ʎ uː˦˨ tʂ] * burma: [b uː˦˨ r m a] * tup: [t̪ uː˦˨ p] * buku: [b u˦˨ k uː] |
|||
Close-Mid |
Occurrences: 16,972 Examples: * elzas: [e˨˦ l z̪ a s̪] * sonde: [s̪ o n d̪ e] * jean: [j e˨˦ aː n] * sferi: [s̪ f e˨˦ r i] Occurrences: 2,029 Examples: * drže: [d̪ r̩˨˦ ʐ eː] * akter: [a˨˦ k t̪ eː r] * space: [s̪ p a˦˨ t̪s̪ eː] * npr: [e n p eː e r] |
Occurrences: 22,565 Examples: * osuda: [o˦˨ s̪ u d̪ a] * carlo: [t̪s̪ a˦˨ r l o] * lopar: [l o˨˦ p aː r] * kenzo: [k e˦˨ n z̪ o] Occurrences: 3,347 Examples: * papom: [p a˦˨ p oː m] * mojih: [m oː˨˦ j i x] * tore: [t̪ oː˦˨ r e] * vagon: [v a˨˦ ɡ oː n] |
|||
Open-Mid |
|||||
Open |
Occurrences: 24,563 Examples: * ladan: [l aː˨˦ d̪ a n] * magma: [m a˨˦ ɡ m a] * osuda: [o˦˨ s̪ u d̪ a] * priča: [p r iː˦˨ tʂ a] Occurrences: 3,936 Examples: * jean: [j e˨˦ aː n] * lopar: [l o˨˦ p aː r] * aras: [a˨˦ r aː s̪] * uzdah: [u˦˨ z̪ d̪ aː x] |
Tones#
˦˨
˨˦