Absolutely delighted & grateful to have 3 contributions (2 at #INTERSPEECH2023, 1 at #SSW12) this year toward TTS for low-resource languages (LRLs). Huge thanks to my supervisors and co-authors @mlcoler @JelskeD @klabbers_esther! Preprints if you're interested:
1. https://t.co/2L67wYjNYO proposes alternatives in a lack of pronunciation dictionaries
2. https://t.co/5oA9nOyFHl expriments with fine-tuning a MOS prediction model focusing on LRLs
3. https://t.co/nEIO0TARtJ investigates strategies to optimize efficiency in TL for TTS in LRLs
Discover the Future of Speech Technology: Join Our Summer School in May!
🤖 Speech Tech Summer School: Speech tech for robots, video games, voice cloning ... and more!
🗓 May 14-17
📍 @CampusFryslan (@univgroningen) the Netherlands
⏰ Space is limited! https://t.co/WWUcLedwNI
Op de dei fan #EDL, @phatt_do fan @CampusFryslan by gearkomste Shaping the Future of Linguistic Diversity:"For minority languages like Frisian, voice technology brings not only a risk, but also an opportunity to increase their presence in everyday life, especially for the youth."
@ashaitar@duolingo Reminds me of the Dutch Duolingo's "Have you talked to the important rhinoceros?" and "The turtle walks past the temple everyday." 😅
Happy to share a new data set for improving language diversity in #NLProc! Our paper "Writing System and Speaker Metadata for 2,800+ Language Varieties", being presented today at #LREC2022, describes our dataset, which lives at https://t.co/EevqU7WHS7. 1/9
Happy to find a few photos of me presenting at @sigul2022#SIGUL2022. Thank you @RoshnaOmer!😀
Enjoyed the opportunity to share my work and really appreciated the questions and discussions that followed!
The third talk at #SIGUL2022
Text-to-Speech for Under-Resourced Languages: Phoneme Mapping and Source Language Selection in Transfer Learning
Phat Do, Matt Coler, Jelske Dijkstra and Esther Klabbers
Today at @sigul2022 (@lrec2022 workshop) I will present my latest work with co-authors @klabbers_esther @JelskeD @mlcoler about phoneme mapping and source-language selection in transfer learning for TTS in under-resourced languages. Looking forward to sharing and discussing!
Thank you @ukrantgroningen for the opportunity to share and discuss my latest work involving a #Frisian synthetic voice, part of my PhD project in synthetic speech for low-resource languages at @CampusFryslan with co-supervisors @klabbers_esther @JelskeD and @mlcoler.
Our PhD researcher @phatt_do is featured in the UKrant for his work on a Frisian synthetic voice. Read all about it! @klabbers_esther @JelskeD @mlcoler
https://t.co/oaTZvb5mjZ
@Flinter_M@GCSCS_RuG (Sorry for replying in English!) Thank you! I've checked it again just now and it seems to be in order, so probably it was just the case that some samples indeed sound very similar to each other (but not by mistake). Anyway, I'll DM you to try and check further if that's okay!
If you speak Frisian, please kindly help with a 20-minute listening experiment of synthetic (artificial) Frisian speech at https://t.co/wEJnfymuXI. Your evaluation of these synthetic Frisian sentences will help assessing different methods to create a Frisian voice. Tige tank!