Interesting links, 25/08/2025
Misc. interesting things.
- English phonetics
- CUBE/Youglish
- Irish
- Lexicon adaptation papers
- Dependency tree editing
- TTS
- ASR
- Hungarian
English phonetics
Why these English phonetic symbols are all WRONG
Lecture 3 The vowel system; clipping
CUBE/Youglish
Current British English pronunciation dictionary
Irish
The Irish of Iorras Aithneach, County Galway
Lexicon adaptation papers
Pronunciation modeling and lexicon adaptation
Fosler-Lussier, E., W. Byrne, and D. Jurafsky. “Pronunciation Modeling and Lexicon Adaptation.” Speech communication 46.2 (2005).
An international English speech corpus for longitudinal study of accent development
Pronunciation modeling for speech technology
A corpus-based study of English pronunciation variations
Pronunciation dependent language models
Dynamic Pronunciation Models for Automatic Speech Recognition
Articulatory feature-based pronunciation modeling
Dependency tree editing
DepEdit — a tool for manipulating dependency trees
udon2/udon2 — A package for manipulating Universal Dependencies trees
DatabaseGroup/apted — APTED algorithm for the Tree Edit Distance
TTS
Probabilistic Speech & Motion Synthesis: Towards More Expressive and Multimodal Generative Models — Shivam’s thesis.
The SIWIS French Speech Synthesis Database
ASR
Pronunciation modeling for large vocabulary conversational speech recognition
@inproceedings{ma98_icslp,
title = {Pronunciation modeling for large vocabulary conversational speech recognition},
author = {Kristine Ma and George Zavaliagkos and Rukmini Iyer},
year = {1998},
booktitle = {5th International Conference on Spoken Language Processing (ICSLP 1998)},
pages = {paper 0866},
doi = {10.21437/ICSLP.1998-655},
issn = {2958-1796},
}
As was found by other researchers, over-generating multiple pronunciations in the dictionary increases word confusability during recognition, often nullifying the advantages of modeling pronunciation variability.
Phonological level wav2vec2-based Mispronunciation Detection and Diagnosis method
@article{shahin2025phonological,
title = {Phonological level wav2vec2-based Mispronunciation Detection and Diagnosis method},
journal = {Speech Communication},
volume = {173},
pages = {103249},
year = {2025},
issn = {0167-6393},
doi = {https://doi.org/10.1016/j.specom.2025.103249},
author = {Mostafa Shahin and Julien Epps and Beena Ahmed},
}
Hungarian
Resources for Learning Hungarian
Triads of movement in Hungarian
A Beginner-Friendly Guide to Hungarian Verb Conjugation