Thai MFA dictionary v2.0.0a#
@techreport{mfa_thai_mfa_dictionary_2022,
author={McAuliffe, Michael and Sonderegger, Morgan},
title={Thai MFA dictionary v2.0.0a},
address={\url{https://mfa-models.readthedocs.io/pronunciation dictionary/Thai/Thai 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 thai_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 Thai transcripts.
This dictionary uses the MFA phone set for Thai, and was used in training the Thai 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 |
Palatal |
Velar |
Glottal |
---|---|---|---|---|---|---|
Nasal |
Occurrences: 9,504 Examples: * สมาคม: [s a ˨ ˩ m a ː ˧ k ʰ o ˧ m] * มินต์: [m i ˦ ˥ n] * ตูม: [t u ː ˧ m] * สมณะ: [s a ˨ ˩ m a ˦ ˥ n a ˦ ˥ ʔ] |
Occurrences: 16,010 Examples: * นิทรา: [n i ˦ ˥ t ̚ t ʰ r a ː ˧] * บรรดา: [b a ˧ n d a ː ˧] * ธนู: [t ʰ a ˦ ˥ n u ː ˧] * หิน: [h i ˩ ˩ ˦ n] |
Occurrences: 6,346 Examples: * บ่วง: [b u a ˨ ˩ ŋ] * น่อง: [n ɔ ˥ ˩ ŋ] * ยิ่ง: [j i ˥ ˩ ŋ] * ละออง: [l a ˦ ˥ ʔ ɔ ː ˧ ŋ] |
|||
Stop |
Occurrences: 2,973 Examples: * เปล: [p l e ː ˧] * เพลซ: [p ʰ l a ˧ j s o ː ˧] * ภาพ: [p ʰ a ː ˥ ˩ p ̚] * ผับ: [p ʰ a ˨ ˩ p ̚] Occurrences: 2,659 Examples: * บีบ: [b i ː ˨ ˩ p ̚] * บอก: [b ɔ ː ˨ ˩ k ̚] * บิทา: [b i ˨ ˩ t ʰ a ː ˧] * บาหลี: [b a ː ˧ l i ː ˩ ˩ ˦] |
Occurrences: 3,808 Examples: * ทมยัน: [t ʰ a ˦ ˥ m a ˦ ˥ j a ˧ n] * วิิถี: [w i ˦ ˥ i ˦ ˥ t ʰ i ː ˩ ˩ ˦] * ตู่: [t u ː ˥ ˩] * จำนง: [t ɕ a ˧ m n o ˧ ŋ] Occurrences: 2,635 Examples: * ด่่วน: [d ɛ ː ˨ ˩ u a ˨ ˩ n] * ระดับ: [r a ˦ ˥ d a ˨ ˩ p ̚] * สนดี: [s a ˨ ˩ n d i ː ˧] * แขกดำ: [k ʰ ɛ ː ˨ ˩ k ̚ d a ˧ m] |
Occurrences: 5,826 Examples: * คณิกา: [k ʰ a ˦ ˥ n i ˦ ˥ k a ː ˧] * กอ.: [k a ˨ ˩ ʔ ɔ ː ˧] * ไอคอน: [ʔ a ˧ j k ʰ ɔ ː ˧ n] * กระจก: [k r a ˨ ˩ t ɕ o ˨ ˩ k ̚] |
Occurrences: 5,131 Examples: * ขนแกะ: [k ʰ o ˩ ˩ ˦ n k ɛ ˨ ˩ ʔ] * เอ๊: [ʔ o ː ˦ ˥] * ลามะ: [l a ː ˧ m a ˦ ˥ ʔ] * แทะ: [t ʰ ɛ ˦ ˥ ʔ] |
||
Affricate |
Occurrences: 2,624 Examples: * ชายคา: [t ɕ ʰ a ː ˧ j k ʰ a ː ˧] * เจ้อ: [t ɕ ɤ ː ˥ ˩] * คิชฌะ: [k ʰ i ˦ ˥ t ̚ t ɕ ʰ a ˦ ˥ ʔ] * ชี่: [t ɕ ʰ i ː ˥ ˩] |
|||||
Sibilant |
Occurrences: 8,736 Examples: * หัวใส: [h u a ˩ ˩ ˦ s a ˩ ˩ ˦ j] * ราศี: [r a ː ˧ s i ː ˩ ˩ ˦] * สังข์: [s a ˩ ˩ ˦ ŋ] * สิ่่ง: [s i ˥ ˩ ɛ ˨ ˩ ŋ] |
|||||
Fricative |
Occurrences: 1,107 Examples: * เจฟ: [t ɕ e ː ˨ ˩ f] * กาแฟน: [k a ː ˧ f ɛ ː ˧ n] * ฟิว: [f i ˧ w] * ฝัง: [f a ˩ ˩ ˦ ŋ] |
Occurrences: 2,069 Examples: * ฮาวาย: [h a ː ˧ w a ː ˧ j] * ห้วงน: [h u a ˥ ˩ ŋ n] * หืน: [h ɯ ː ˧ n] * ดาหาร: [d a ː ˧ h a ː ˩ ˩ ˦ n] |
||||
Approximant |
Occurrences: 5,593 Examples: * วาท: [w a ː ˥ ˩ t ̚] * เจัา: [t ɕ a ˧ w] * ละว้า: [l a ˦ ˥ w a ː ˦ ˥] * เวส: [w e ˦ ˥ s] |
Occurrences: 7,309 Examples: * ไยอบ: [j ɔ ː ˥ ˩ p ̚] * เหยง: [j o ˧ ŋ] * เยิ้ม: [j ɤ ː ˦ ˥ m] * เฟือย: [f ɯ a ˧ j] |
||||
Trill |
Occurrences: 8,856 Examples: * ตรีส์: [t r i ː ˧ s] * อำระ: [ʔ a ˧ m r a ˦ ˥ ʔ] * พระวจ: [p ʰ r a ˦ ˥ w t ̚] * พรุ่ง: [p ʰ r u ˥ ˩ ŋ] |
|||||
Lateral |
Occurrences: 5,932 Examples: * เลย.: [l e ː ˧ j ɔ ː ˧] * ลียัง: [l i ː ˧ j a ˧ ŋ] * แลคโต: [l ɛ ˦ ˥ k ̚ t o ː ˧] * ปล้อง: [p l ɔ ˥ ˩ ŋ] |
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: 6,842 Examples: * ชาดี: [t ɕ ʰ a ː ˧ d i ː ˧] * สิิน: [s i ˨ ˩ ɤ ː ˧ n] * วิถี: [w i ˦ ˥ t ʰ i ː ˩ ˩ ˦] * เหริน: [h e ː ˩ ˩ ˦ r i ˧ n] Occurrences: 4,158 Examples: * มราฐี: [m a ˦ ˥ r a ː ˧ t ʰ i ː ˩ ˩ ˦] * หมอผี: [m ɔ ː ˩ ˩ ˦ p ʰ i ː ˩ ˩ ˦] * บี่: [b i ː ˨ ˩] * คอตีบ: [k ʰ ɔ ː ˧ t i ː ˨ ˩ p ̚] |
Occurrences: 625 Examples: * ซึ้ง: [s ɯ ˦ ˥ ŋ] * คัตสึ: [k ʰ a ˦ ˥ t ̚ s ɯ ˨ ˩ ʔ] * ดูถึง: [d u ː ˧ t ʰ ɯ ˩ ˩ ˦ ŋ] * มึม: [m ɯ ˧ m] Occurrences: 734 Examples: * ตื๊อ: [t ɯ ː ˦ ˥] * รื่น: [r ɯ ː ˥ ˩ n] * ฝืน: [f ɯ ː ˩ ˩ ˦ n] * สื่ง: [s ɯ ː ˨ ˩ e ˨ ˩ ŋ] Occurrences: 3,498 Examples: * หนุน: [n u ˩ ˩ ˦ n] * โชยุ: [t ɕ ʰ o ː ˧ j u ˨ ˩ ʔ] * นวล: [n u a ˧ n] * บุ้ง: [b u ˥ ˩ ŋ] Occurrences: 2,364 Examples: * หูย: [h u ː ˩ ˩ ˦ j] * รู้จบ: [r u ː ˦ ˥ t ɕ o ˨ ˩ p ̚] * นู่น: [n u ː ˩ ˩ ˦ ɛ ˥ ˩ n] * ขู่: [k ʰ u ː ˨ ˩] |
|||
Close-Mid |
Occurrences: 1,755 Examples: * ไลเคน: [l a ˧ j k ʰ e ˥ ˩ n] * เม่น: [m e ˥ ˩ n] * ราเชล: [r a ː ˧ t ɕ ʰ e ˧ l] * เซ็ง: [s e ˧ ŋ] Occurrences: 2,633 Examples: * เขน: [k ʰ e ː ˩ ˩ ˦ n] * เรณู: [r e ː ˧ n u ː ˧] * เดวิด: [d e ː ˧ w i ˨ ˩ t ̚] * เวฟ: [w e ː ˦ ˥ p ̚] |
Occurrences: 222 Examples: * เดิ้ล: [d ɤ ˥ ˩ n] * เสย: [s ɤ ˩ ˩ ˦ j] * เบอระ: [b ɤ ː ˧ r a ˦ ˥ ʔ] * เคิ่ล: [k ʰ ɤ ˥ ˩ l] Occurrences: 1,597 Examples: * เพ้อ: [p ʰ ɤ ː ˦ ˥] * เชิง: [t ɕ ʰ ɤ ː ˧ ŋ] * เตอร์: [t ɤ ː ˥ ˩] * เหม่อ: [m ɤ ː ˨ ˩] Occurrences: 3,709 Examples: * ทนทาน: [t ʰ o ˧ n t ʰ a ː ˧ n] * ทูต้ม: [t ʰ u ː ˧ t o ˥ ˩ m] * รสชาด: [r o ˦ ˥ t ̚ t ɕ ʰ a ː ˥ ˩ t ̚] * หกสิบ: [h o ˨ ˩ k ̚ s i ˨ ˩ p ̚] Occurrences: 2,450 Examples: * เลโก้: [l o ː ˧ k o ː ˥ ˩] * โอ๊ต: [ʔ o ː ˦ ˥ t ̚] * โหว่: [w o ː ˨ ˩] * โค๊ก: [k ʰ o ː ˦ ˥ k ̚] |
|||
Open-Mid |
Occurrences: 753 Examples: * แข็ง: [k ʰ ɛ ˩ ˩ ˦ ŋ] * แม่ง: [m ɛ ˥ ˩ ŋ] * แต๋ว: [t ɛ ˩ ˩ ˦ w] * ไรฟะ: [r a ˧ j f ɛ ˨ ˩ ʔ] Occurrences: 1,922 Examples: * แฮท: [h ɛ ː ˦ ˥ t ̚] * แววาว: [w ɛ ː ˧ w a ː ˧ w] * ดูแย่: [d u ː ˧ j ɛ ː ˥ ˩] * แค่ย: [k ʰ ɛ ː ˥ ˩ j] |
Occurrences: 1,316 Examples: * คอม: [k ʰ ɔ ˧ m] * หอย: [h ɔ ˩ ˩ ˦ j] * อะตอม: [ʔ a ˨ ˩ t ɔ ˧ m] * สก็อต: [s a ˨ ˩ k ɔ ˨ ˩ t ̚] Occurrences: 4,591 Examples: * พอก: [p ʰ ɔ ː ˥ ˩ k ̚] * ก็ไอ: [k ɔ ː ˥ ˩ ʔ a ˧ j] * สนง.: [s ɔ ː ˩ ˩ ˦ n ɔ ː ˧ ŋ ɔ ː ˧] * กอล์ฟ: [k ɔ ː ˦ ˥ f] |
|||
Open |
Occurrences: 21,999 Examples: * เชือด: [t ɕ ʰ ɯ a ˥ ˩ t ̚] * ขบวร: [k ʰ a ˨ ˩ b u a ˧ n] * สกุณ: [s a ˨ ˩ k u ˧ n] * สารท: [s a ː ˩ ˩ ˦ r a ˦ ˥ t ʰ a ˦ ˥] Occurrences: 14,324 Examples: * ผีฟ้า: [p ʰ i ː ˩ ˩ ˦ f a ː ˦ ˥] * น้ำตก: [n a ː ˦ ˥ m t o ˨ ˩ k ̚] * สต๊าฟ: [s a ˨ ˩ t a ː ˦ ˥ f] * ชาด: [t ɕ ʰ a ː ˥ ˩ t ̚] |
Diphthongs#
ia
ua
ɯa
Tones#
˥˩
˦˥
˧
˨˩
˩˩˦