Ctrl+K
Ctrl+K

Portuguese (Brazil) MFA dictionary v2.0.0a#

@techreport{mfa_portuguese_brazil_mfa_dictionary_2022,
	author={McAuliffe, Michael and Sonderegger, Morgan},
	title={Portuguese (Brazil) MFA dictionary v2.0.0a},
	address={\url{https://mfa-models.readthedocs.io/pronunciation dictionary/Portuguese/Portuguese (Brazil) MFA dictionary v2_0_0a.html}},
	year={2022},
	month={May},
}

G2P models
../../_images/full_logo_yellow.svg

Installation#

Install from the MFA command line:

mfa model download dictionary portuguese_brazil_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 Portuguese transcripts.

This dictionary uses the MFA phone set for Portuguese, and was used in training the Portuguese 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

Nasal
Occurrences:
7,034
Examples:
* milhã:
[m i ʎ ɐ ̃]
* melo:
[m e l u]
* motim:
[m o t ʃ i ̃]
* mesmo:
[m e m u]
Occurrences:
4,091
Examples:
* menem:
[m e n e ̃ j ̃]
* tênue:
[t e n w i]
* numa:
[n u m ɐ]
* turno:
[t u x n u]
Occurrences:
1,513
Examples:
* tony:
[t o ɲ i]
* lenin:
[l e ɲ i ̃]
* cisne:
[s i z ɲ i]
* anita:
[ɐ ɲ i t ɐ]

Stop
Occurrences:
7,132
Examples:
* pedra:
[p ɛ d ɾ ɐ]
* pilar:
[p i l a x]
* iapu:
[j a p u]
* bope:
[b ɔ p i]
Occurrences:
4,102
Examples:
* psdb:
[p e e s d e b e]
* bein:
[b e ̃ j ̃]
* benta:
[b e ̃ t a]
* ibiam:
[i b i ɐ ̃ w ̃]
Occurrences:
10,437
Examples:
* botox:
[b o t o k s]
* tirar:
[t ʃ i ɾ a x]
* ponto:
[p o ̃ t o]
* monte:
[m o ̃ t e]
Occurrences:
8,774
Examples:
* andir:
[ɐ ̃ d ʒ i x]
* idoso:
[i d o z u]
* dusk:
[d u s k]
* pende:
[p e ̃ d ʒ i]
Occurrences:
1,064
Examples:
* caém:
[c e ̃ j ̃]
* makye:
[m a c i]
* york:
[j ɔ x c i]
* kevin:
[c ɛ v i ̃]
Occurrences:
408
Examples:
* gueto:
[ɟ ɛ t u]
* gate:
[ɟ e j t ʃ]
* magno:
[m a ɟ i n u]
* jung:
[ʒ u ̃ ɟ i]
Occurrences:
8,737
Examples:
* copa:
[k ɔ p ɐ]
* sulco:
[s u w k u]
* curar:
[k u ɾ a x]
* caros:
[k a ɾ u s]
Occurrences:
2,975
Examples:
* magos:
[m a ɡ u s]
* ong:
[o ̃ ɡ]
* logos:
[l o ɡ u s]
* lugar:
[l u ɡ a x]

Affricate
Occurrences:
4,032
Examples:
* ótico:
[ɔ t ʃ i k u]
* patis:
[p a t ʃ i s]
* net:
[n ɛ t ʃ i]
* note:
[n o w t ʃ]
Occurrences:
2,486
Examples:
* wood:
[v o u d ʒ]
* disto:
[d ʒ i s t o]
* dieta:
[d ʒ i e t ɐ]
* david:
[d e j v i d ʒ]

Sibilant
Occurrences:
19,445
Examples:
* algas:
[a w ɡ ɐ s]
* secas:
[s ɛ k ɐ s]
* estas:
[ɛ s t ɐ s]
* tamuz:
[t ɐ ̃ u j s]
Occurrences:
3,476
Examples:
* trazê:
[t ɾ a z e]
* casé:
[k a z ɛ]
* idoso:
[i d o z u]
* ganzo:
[ɡ ɐ ̃ z u]
Occurrences:
1,780
Examples:
* encha:
[e ̃ ʃ ɐ]
* rocha:
[x ɔ ʃ ɐ]
* ente:
[e ̃ t ʃ i]
* deixa:
[d e j ʃ ɐ]
Occurrences:
2,099
Examples:
* judeu:
[ʒ u d e w]
* old:
[ɔ l d ʒ]
* orgia:
[o x ʒ i a]
* beije:
[b e j ʒ e]

Fricative
Occurrences:
3,150
Examples:
* fogo:
[f o ɡ u]
* falta:
[f a w t ɐ]
* tufão:
[t u f ɐ ̃ w ̃]
* fibra:
[f i b ɾ ɐ]
Occurrences:
4,177
Examples:
* envia:
[e ̃ v i a]
* ferva:
[f ɛ x v ɐ]
* njiva:
[ʒ i v ɐ]
* vaz:
[v a s]

Approximant
Occurrences:
4,913
Examples:
* joão:
[ʒ o ɐ ̃ w ̃]
* elói:
[ɛ w ɔ j]
* álbum:
[a w b u ̃]
* joao:
[ʒ w a w]
Occurrences:
2,231
Examples:
* sazão:
[s a z ɐ ̃ w ̃]
* mamão:
[m ɐ m ɐ ̃ w ̃]
* duram:
[d u ɾ ɐ ̃ w ̃]
* cotão:
[k o t ɐ ̃ w ̃]
Occurrences:
4,544
Examples:
* fúria:
[f u ɾ j ɐ]
* dylan:
[d a j l ɐ ̃]
* miami:
[m a j ɐ m i]
* ferem:
[f ɛ ɾ e ̃ j ̃]
Occurrences:
1,792
Examples:
* bein:
[b e ̃ j ̃]
* caém:
[c e ̃ j ̃]
* vivem:
[v i v e ̃ j ̃]
* cabem:
[k a b e ̃ j ̃]

Tap
Occurrences:
12,891
Examples:
* erick:
[e ɾ i k]
* grega:
[ɡ ɾ e ɡ ɐ]
* crie:
[k ɾ i]
* pras:
[p ɾ a s]

Lateral
Occurrences:
4,620
Examples:
* close:
[k l o z i]
* telos:
[t e l u s]
* luck:
[l u k]
* lady:
[l a d i]
Occurrences:
2,923
Examples:
* álibi:
[a ʎ i b i]
* alho:
[a ʎ u]
* clico:
[k ʎ i k u]
* linus:
[ʎ i n u s]

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:
23,708
Examples:
* laíse:
[l a i s i]
* jaine:
[ʒ a j ɲ i]
* til:
[t ʃ i w]
* ave:
[a v i]
Occurrences:
14,318
Examples:
* ósseo:
[ɔ s e u]
* motos:
[m ɔ t u s]
* cedo:
[s e d u]
* cairo:
[k a j ɾ u]

Close-Mid
Occurrences:
14,339
Examples:
* tower:
[t ɔ a w e x]
* agem:
[a ʒ e ̃ j ̃]
* cemig:
[s e m i ɡ]
* kylie:
[k a j ʎ e]
Occurrences:
9,555
Examples:
* torna:
[t o x n ɐ]
* movem:
[m o v e ̃ j ̃]
* poró:
[p o ɾ ɔ]
* elton:
[ɛ w t o ̃]

Open-Mid
Occurrences:
3,191
Examples:
* hebe:
[ɛ b i]
* sénic:
[s ɛ ɲ i k]
* pedra:
[p ɛ d ɾ ɐ]
* ibaté:
[i b a t ɛ]
Occurrences:
2,113
Examples:
* óvulo:
[ɔ v u l u]
* dobre:
[d ɔ b ɾ i]
* godói:
[ɡ o d ɔ j]
* roda:
[x ɔ d ɐ]
Occurrences:
9,939
Examples:
* falha:
[f a ʎ ɐ]
* baba:
[b a b ɐ]
* cega:
[s ɛ ɡ ɐ]
* adria:
[a d ɾ i ɐ]

Open
Occurrences:
24,243
Examples:
* base:
[b a z i]
* major:
[m a ʒ ɔ x]
* dali:
[d a ʎ i]
* lidei:
[l a j d ʒ e j]

Nasal Vowels#

Front

Near-Front

Central

Near-Back

Back

Close
Occurrences:
2,808
Examples:
* ruim:
[x u i ̃]
* tinha:
[t ʃ i ̃ j ̃ a]
* linda:
[ʎ i ̃ d ɐ]
* linha:
[ʎ i ̃ j ̃ ɐ]
Occurrences:
536
Examples:
* junta:
[ʒ u ̃ t ɐ]
* junho:
[ʒ u ̃ j ̃ u]
* sunga:
[s u ̃ ɡ ɐ]
* unhar:
[u ̃ j ̃ a ɾ]

Close-Mid
Occurrences:
4,035
Examples:
* nuvem:
[n u v e ̃ j ̃]
* nem:
[n e ̃ j ̃]
* open:
[o p e ̃]
* vivem:
[v i v e ̃ j ̃]
Occurrences:
2,433
Examples:
* édson:
[ɛ d s o ̃]
* leões:
[l e o ̃ j ̃ s]
* jhon:
[ʒ o ̃]
* monte:
[m o ̃ t e]

Open-Mid
Occurrences:
5,513
Examples:
* manco:
[m ɐ ̃ k u]
* rolam:
[x o l ɐ ̃ w ̃]
* ansar:
[ɐ ̃ z a x]
* cante:
[k ɐ ̃ t ʃ i]

Open
On this page
Show Source