Swahili MFA dictionary v2.0.0a#
@techreport{mfa_swahili_mfa_dictionary_2022,
author={McAuliffe, Michael and Sonderegger, Morgan},
title={Swahili MFA dictionary v2.0.0a},
address={\url{https://mfa-models.readthedocs.io/pronunciation dictionary/Swahili/Swahili 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 swahili_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 Swahili transcripts.
This dictionary uses the MFA phone set for Swahili, and was used in training the Swahili 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: 13,053 Examples: * mteja: [m t ɛ d ʒ ɑ] * kombo: [k ɔ m b ɔ] * mtu: [m t u] * mkasa: [m k ɑ s ɑ] |
Occurrences: 219 Examples: * mfano: [ɱ f ɑ n ɔ] * mfike: [ɱ f i k ɛ] * mfugo: [ɱ f u ɠ ɔ] * mvezo: [ɱ v ɛ z ɔ] |
Occurrences: 14,657 Examples: * duni: [d u n i] * nan: [n ɑ n] * tian: [t i ɑ n] * twin: [t w i n] |
Occurrences: 1,484 Examples: * nyali: [ɲ ɑ l i] * anywe: [ɑ ɲ w ɛ] * nyayo: [ɲ ɑ j ɔ] * nyada: [ɲ ɑ ɗ ɑ] |
Occurrences: 3,735 Examples: * yanga: [j ɑ ŋ ɡ ɑ] * mungu: [m u ŋ ɡ u] * tungo: [t u ŋ ɡ ɔ] * hangi: [h ɑ ŋ ɡ i] |
|||
Stop Plain |
Occurrences: 4,375 Examples: * upana: [u p ɑ n ɑ] * meupe: [m ɛ u p ɛ] * chips: [t ʃ i p s] * tipp: [t i p] Occurrences: 2,452 Examples: * lambo: [l ɑ m b ɔ] * bona: [b ɔ n ɑ] * babel: [b ɑ ɓ ɛ l] * busu: [b u s u] |
Occurrences: 10,788 Examples: * vocha: [v ɔ t ʃ ɑ] * grant: [ɡ ɾ ɑ n t] * fiti: [f i t i] * uchu: [u t ʃ u] Occurrences: 2,670 Examples: * jumbe: [d ʒ u m b ɛ] * wjia: [w d ʒ i ɑ] * gwiji: [ɡ w i d ʒ i] * mkojo: [m k ɔ d ʒ ɔ] |
Occurrences: 18,250 Examples: * tamko: [t ɑ m k ɔ] * kitui: [k i t u i] * kwako: [k w ɑ k ɔ] * kede: [k ɛ ɗ ɛ] Occurrences: 3,968 Examples: * wingi: [w i ŋ ɡ i] * grau: [ɡ ɾ ɑ u] * gates: [ɡ ɑ t ɛ s] * gundi: [ɡ u n d i] |
|||||
Implosive |
Occurrences: 2,697 Examples: * ajibu: [ɑ d ʒ i ɓ u] * vibao: [v i ɓ ɑ ɔ] * aruba: [ɑ ɾ u ɓ ɑ] * labre: [l ɑ ɓ ɾ ɛ] |
Occurrences: 1,742 Examples: * abdil: [ɑ ɓ ɗ i l] * wadoe: [w ɑ ɗ ɔ ɛ] * kadri: [k ɑ ɗ ɾ i] * dada: [d ɑ ɗ ɑ] |
Occurrences: 1,786 Examples: * soglo: [s ɔ ɠ l ɔ] * zigi: [z i ɠ i] * njuga: [n d ʒ u ɠ ɑ] * agca: [ɑ ɠ k] |
|||||
Affricate |
Occurrences: 2,025 Examples: * uchao: [u t ʃ ɑ ɔ] * chafu: [t ʃ ɑ f u] * nchi: [n t ʃ i] * nchii: [n t ʃ i ː] Occurrences: 3,463 Examples: * mtaji: [m t ɑ d ʒ i] * jimmy: [d ʒ i m i] * hajui: [h ɑ d ʒ u i] * jama: [d ʒ ɑ m ɑ] |
|||||||
Sibilant |
Occurrences: 5,493 Examples: * plus: [p l u s] * msije: [m s i d ʒ ɛ] * snake: [s n ɛ k] * hasi: [h ɑ s i] Occurrences: 6,484 Examples: * zumbu: [z u m b u] * mzima: [m z i m ɑ] * nguzo: [ŋ ɡ u z ɔ] * ziada: [z i ɑ ɗ ɑ] |
Occurrences: 4,283 Examples: * kushi: [k u ʃ i] * shaba: [ʃ ɑ ɓ ɑ] * flash: [f l ɑ ʃ] * roush: [ɾ ɔ u ʃ] |
||||||
Fricative |
Occurrences: 3,394 Examples: * dufu: [d u f u] * vifo: [v i f ɔ] * uefa: [u ɛ f ɑ] * chifu: [t ʃ i f u] Occurrences: 2,216 Examples: * vipi: [v i p i] * wivu: [w i v u] * vyeti: [v j ɛ t i] * glavu: [ɡ l ɑ v u] |
Occurrences: 315 Examples: * the: [θ ɛ] * edith: [ɛ ɗ i θ] * youth: [j ɔ u θ] * this: [θ i s] Occurrences: 581 Examples: * dhidi: [ð i ɗ i] * riyad: [ɾ i j ɑ ð ɑ] * dhabi: [ð ɑ ɓ i] * idhaa: [i ð ɑ ː] |
Occurrences: 4,883 Examples: * aaah: [ɑ ː h] * hold: [h ɔ l d] * hawa: [h ɑ w ɑ] * hoi: [h ɔ i] |
|||||
Approximant |
Occurrences: 13,299 Examples: * kuwa: [k u w ɑ] * iwapo: [i w ɑ p ɔ] * wiki: [w i k i] * maswa: [m ɑ s w ɑ] |
Occurrences: 3,947 Examples: * lyon: [l j ɔ n] * eyasi: [ɛ j ɑ s i] * yani: [j ɑ n i] * yusuf: [j u s u f] |
||||||
Tap |
Occurrences: 5,303 Examples: * rocky: [ɾ ɔ k i] * rico: [ɾ i k ɔ] * chura: [t ʃ u ɾ ɑ] * pira: [p i ɾ ɑ] |
|||||||
Lateral |
Occurrences: 13,972 Examples: * leta: [l ɛ t ɑ] * alafu: [ɑ l ɑ f u] * iliki: [i l i k i] * mlete: [m l ɛ t ɛ] |
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: 43,788 Examples: * beila: [b ɛ i l ɑ] * silas: [s i l ɑ s] * tuki: [t u k i] * salim: [s ɑ l i m] Occurrences: 520 Examples: * utii: [u t i ː] * bidii: [b i ɗ i ː] * miiba: [m i ː ɓ ɑ] * nabii: [n ɑ ɓ i ː] |
Occurrences: 21,892 Examples: * cum: [k u m] * tatau: [t ɑ t ɑ u] * uume: [u ː m ɛ] * binu: [b i n u] Occurrences: 291 Examples: * kuupa: [k u ː p ɑ] * nafuu: [n ɑ f u ː] * miguu: [m i ɠ u ː] * guu: [ɡ u ː] |
|||
Close-Mid |
|||||
Open-Mid |
Occurrences: 15,798 Examples: * gesi: [ɡ ɛ s i] * jeff: [d ʒ ɛ f] * wakae: [w ɑ k ɑ ɛ] * imler: [i m l ɛ ɾ] Occurrences: 172 Examples: * leeds: [l ɛ ː ɗ s] * mizee: [m i z ɛ ː] * njee: [n d ʒ ɛ ː] * vudee: [v u ɗ ɛ ː] |
Occurrences: 12,413 Examples: * wolff: [w ɔ l f] * jaluo: [d ʒ ɑ l u ɔ] * tupo: [t u p ɔ] * kuona: [k u ɔ n ɑ] Occurrences: 213 Examples: * etoo: [ɛ t ɔ ː] * moon: [m ɔ ː n] * akyoo: [ɑ k j ɔ ː] * noor: [n ɔ ː ɾ] |
|||
Open |
Occurrences: 63,025 Examples: * maili: [m ɑ i l i] * vyao: [v j ɑ ɔ] * kanza: [k ɑ n z ɑ] * style: [s t ɑ j l] Occurrences: 1,230 Examples: * tamaa: [t ɑ m ɑ ː] * kuvaa: [k u v ɑ ː] * kavaa: [k ɑ v ɑ ː] * gaal: [ɡ ɑ ː l] |