Thai MFA dictionary v2.0.0#
@techreport{mfa_thai_mfa_dictionary_2022,
author={McAuliffe, Michael and Sonderegger, Morgan},
title={Thai MFA dictionary v2.0.0},
address={\url{https://mfa-models.readthedocs.io/pronunciation dictionary/Thai/Thai 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 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: 9,141 Examples: * โมเพด: [m oː˧ pʰ e˨˩ t̚] * พิมพ์: [pʰ i˧ m] * กำแพง: [k a˧ m pʰ ɛː˧ ŋ] * เมีย: [m ia˧] |
Occurrences: 15,369 Examples: * ลักษณ: [l a˦˥ k̚ s a˧ n] * เฮียน: [h ia˧ n] * ฟินไป: [f i˧ n p a˧ j] * สิิ้น: [s i˨˩ i˥˩ n] |
Occurrences: 6,147 Examples: * พงศ์: [pʰ o˧ ŋ] * กำแพง: [k a˧ m pʰ ɛː˧ ŋ] * งีบ: [ŋ iː˥˩ p̚] * ลงทุน: [l o˧ ŋ tʰ u˧ n] |
|||
Stop |
Occurrences: 2,838 Examples: * ตีปีก: [t iː˧ p iː˨˩ k̚] * ฟินไป: [f i˧ n p a˧ j] * แปะ: [p ɛ˨˩ ʔ] * ปู้อี: [p uː˥˩ ʔ iː˧] Occurrences: 2,502 Examples: * ระบุ: [r a˦˥ b u˨˩ ʔ] * โบ้: [b oː˥˩] * บีีบ: [b iː˨˩ iː˨˩ p̚] * บ่าย: [b aː˨˩ j] |
Occurrences: 3,617 Examples: * ติด: [t i˨˩ t̚] * ตีปีก: [t iː˧ p iː˨˩ k̚] * ซิตี้: [s i˦˥ t iː˥˩] * ตื่น: [t ɯː˨˩ n] Occurrences: 2,488 Examples: * ราษฎร: [r aː˥˩ t̚ s a˨˩ d ɔː˧ n] * กดดัน: [k o˨˩ t̚ d a˧ n] * การดู: [k aː˧ n d uː˧] * ดั้ง: [d a˥˩ ŋ] |
Occurrences: 5,589 Examples: * กลัด: [k l a˨˩ t̚] * การพบ: [k aː˧ n pʰ o˦˥ p̚] * กำแพง: [k a˧ m pʰ ɛː˧ ŋ] * เกษส.: [k a˨˩ s eː˨˩ s ɔː˩˩˦] |
Occurrences: 4,725 Examples: * ระบุ: [r a˦˥ b u˨˩ ʔ] * อโศก: [ʔ a˨˩ s oː˨˩ k̚] * แปะ: [p ɛ˨˩ ʔ] * ปู้อี: [p uː˥˩ ʔ iː˧] |
||
Affricate |
Occurrences: 2,509 Examples: * คาใจ: [kʰ aː˧ tɕ a˧ j] * จะตาย: [tɕ a˨˩ t aː˧ j] * เจด: [tɕ eː˨˩ t̚] * ไข้ใจ: [kʰ a˥˩ j tɕ a˧ j] |
|||||
Sibilant |
Occurrences: 8,198 Examples: * ลักษณ: [l a˦˥ k̚ s a˧ n] * อโศก: [ʔ a˨˩ s oː˨˩ k̚] * สิิ้น: [s i˨˩ i˥˩ n] * วิสัย: [w i˦˥ s a˩˩˦ j] |
|||||
Fricative |
Occurrences: 962 Examples: * ฟินไป: [f i˧ n p a˧ j] * หูฟัง: [h uː˩˩˦ f a˧ ŋ] * ฟรัง: [f r a˧ ŋ] * ฟััง: [f a˧ ŋ] |
Occurrences: 1,974 Examples: * เฮียน: [h ia˧ n] * เหรา: [h eː˩˩˦ r aː˧] * ห้าม: [h aː˥˩ m] * โรหิต: [r oː˧ h i˨˩ t̚] |
||||
Approximant |
Occurrences: 5,365 Examples: * วิสัย: [w i˦˥ s a˩˩˦ j] * วิฬา: [w i˦˥ l aː˧] * ว่าย: [w aː˥˩ j] * แรเงา: [r ɛː˧ ŋ a˧ w] |
Occurrences: 6,962 Examples: * ฟินไป: [f i˧ n p a˧ j] * วิสัย: [w i˦˥ s a˩˩˦ j] * ไอกิส: [ʔ a˧ j k i˦˥ t̚ s a˨˩] * ว่าย: [w aː˥˩ j] |
||||
Trill |
Occurrences: 8,413 Examples: * ระบุ: [r a˦˥ b u˨˩ ʔ] * เหรา: [h eː˩˩˦ r aː˧] * ราษฎร: [r aː˥˩ t̚ s a˨˩ d ɔː˧ n] * โรหิต: [r oː˧ h i˨˩ t̚] |
|||||
Lateral |
Occurrences: 5,521 Examples: * ลักษณ: [l a˦˥ k̚ s a˧ n] * กลัด: [k l a˨˩ t̚] * วิฬา: [w i˦˥ l aː˧] * กลีบ: [k l iː˨˩ p̚] |
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: 6,518 Examples: * วิฬา: [w i˦˥ l aː˧] * วิ้ง: [w i˥˩ ŋ] * หิวโซ: [h i˩˩˦ w s oː˥˩] * ไอกิส: [ʔ a˧ j k i˦˥ t̚ s a˨˩] Occurrences: 3,947 Examples: * สีทอง: [s iː˩˩˦ tʰ ɔː˧ ŋ] * ซิตี้: [s i˦˥ t iː˥˩] * ชีส: [tɕʰ iː˦˥ s] * ตีปีก: [t iː˧ p iː˨˩ k̚] |
Occurrences: 613 Examples: * จึ๋ย: [tɕ ɯ˩˩˦ j] * ดูถึง: [d uː˧ tʰ ɯ˩˩˦ ŋ] * สะอึก: [s a˨˩ ʔ ɯ˨˩ k̚] * พึง: [pʰ ɯ˧ ŋ] Occurrences: 723 Examples: * ถือสา: [tʰ ɯː˩˩˦ s aː˩˩˦] * ฦๅชา: [l ɯː˧ tɕʰ aː˧] * มืดลง: [m ɯː˥˩ t̚ l o˧ ŋ] * ลืมพก: [l ɯː˧ m pʰ o˦˥ k̚] Occurrences: 3,369 Examples: * ทุม: [tʰ u˧ m] * มุุ่ง: [m u˦˥ u˥˩ ŋ] * ทารุณ: [tʰ aː˧ r u˧ n] * ทุเรศ: [tʰ u˦˥ r eː˥˩ t̚] Occurrences: 2,278 Examples: * งูสุด: [ŋ uː˧ s u˨˩ t̚] * หูฟัง: [h uː˩˩˦ f a˧ ŋ] * พู่: [pʰ uː˥˩] * ปูด: [p uː˨˩ t̚] |
|||
Close-Mid |
Occurrences: 1,617 Examples: * เก่ง: [k e˨˩ ŋ] * เม็ง: [m e˧ ŋ] * เหม็น: [m e˩˩˦ n] * เหม็ด: [m e˦˥ t̚] Occurrences: 2,324 Examples: * เฮด: [h eː˦˥ t̚] * เอ๋: [ʔ eː˩˩˦] * เมษ: [m eː˧ s a˨˩] * ทุเรศ: [tʰ u˦˥ r eː˥˩ t̚] |
Occurrences: 193 Examples: * ไธอะ: [tʰ ɤ˨˩ ʔ] * เฉย: [tɕʰ ɤ˩˩˦ j] * เศิก: [s ɤ˨˩ k̚] * เลอะ: [l ɤ˦˥ ʔ] Occurrences: 1,440 Examples: * ลีมิน: [l iː˧ m ɤː˧ n] * เสนอ: [s a˨˩ n ɤː˩˩˦] * เลิฟ: [l ɤː˥˩ f] * เมอร์: [m ɤː˥˩] Occurrences: 3,575 Examples: * รังนก: [r a˧ ŋ n o˦˥ k̚] * ด้น: [d o˥˩ n] * ต้ม: [t o˥˩ m] * สมมติ: [s o˩˩˦ m m o˦˥ t̚] Occurrences: 2,195 Examples: * โสม: [s oː˩˩˦ m] * โน่: [n oː˥˩] * โซดาอ: [s oː˧ d aː˧ ʔ oː˥˩] * โนติ: [n oː˦˥ t̚] |
|||
Open-Mid |
Occurrences: 692 Examples: * แรบ: [r ɛ˥˩ p̚] * แย่ง: [j ɛ˥˩ ŋ] * แร้ว: [r ɛ˥˩ w] * แทง: [tʰ ɛ˧ ŋ] Occurrences: 1,810 Examples: * แรเงา: [r ɛː˧ ŋ a˧ w] * แล้: [l ɛː˦˥] * แยก: [j ɛː˥˩ k̚] * พิศวง: [pʰ i˦˥ t̚ s a˨˩ w ɛː˩˩˦ ŋ] |
Occurrences: 1,222 Examples: * ช็อป: [tɕʰ ɔ˦˥ p̚] * เซาะ: [s ɔ˦˥ ʔ] * คล่อง: [kʰ l ɔ˥˩ ŋ] * มาก็็: [m aː˧ k ɔː˥˩ ɔ˥˩ ʔ] Occurrences: 4,369 Examples: * ธปท.: [tʰ ɔː˧ p ɔː˧ tʰ ɔː˧] * พ่อตา: [pʰ ɔː˥˩ t aː˧] * ราษฎร: [r aː˥˩ t̚ s a˨˩ d ɔː˧ n] * ปอด: [p ɔː˨˩ t̚] |
|||
Open |
Occurrences: 21,013 Examples: * อ่่าว: [ʔ a˥˩ w a˥˩ w] * กลัด: [k l a˨˩ t̚] * ล่ะ: [l a˥˩ ʔ] * ขัย: [kʰ a˩˩˦ j] Occurrences: 13,936 Examples: * สถาน: [s a˨˩ tʰ aː˩˩˦ n] * นาคี: [n aː˧ kʰ iː˧] * การพบ: [k aː˧ n pʰ o˦˥ p̚] * บาตร: [b aː˨˩ t̚] |
Diphthongs#
ia
ua
ɯa
Tones#
˥˩
˦˥
˧
˨˩
˩˩˦