JAIST-ASEAN Deepfake Detection Hub Symposium and ASEAN IVO Meeting


JAIST-ASEAN Deepfake Voice Detection Hub

JAIST-ASEAN Deepfake Voice Detection Hub aims to create a collaborative research environment and to bring together researchers and academics from JAIST and prominent universities and research institutes in the ASEAN region to work on the development of deepfake voice detection technology in Asian languages.

To foster wider engagement and showcase the project's advancements, we concurrently held a one-day symposium alongside the third meeting of 'ASEAN IVO project on Spoof Detection for Automatic Speaker Verification' at JAIST. This event was attended by all key representatives from Japan, Thailand, Indonesia, Brunei Darussalam, and Myanmar institutions. A total of 30 participants attended the meeting in-person and online.

Hub Members (Representative)

  • Candy Olivia Mawalim (PI) (JAIST, Japan)
  • Masashi Unoki (Co-I) (JAIST, Japan)
  • Shogo Okada (Co-I) (JAIST, Japan)
  • Pakinee Aimmanee (Co-I) (SIIT, Thailand)
  • Waree Kongprawechnon (Co-I) (SIIT, Thailand)
  • Kasorn Galajit (Co-I) (NECTEC, Thailand)
  • Jessada Karnjana (Co-I) (NECTEC, Thailand)
  • Win Pa Pa (Co-I) (UCSY, Myanmar)
  • Xugang Lu (Co-I) (NICT, Japan)
  • Dk Hayati Pg Hj Mohd Yassin (Co-I) (UBD, Brunei Darussalam)
  • Dessi Puji Lestari (Co-I) (ITB, Indonesia)


Research Sharing and Discussion

Day 1: ASEAN-IVO Technical Meeting

The first day focuses on research presentations from different institutions:

Morning Session:
  • Welcome remarks from local host (Prof. Masashi Unoki), ASEAN-IVO project leader (Dr. Kasorn Galajit), and NICT representative (Wang Yan)
  • Presentations from NECTEC (Thailand) on ThaiSpoof improvement and SIIT on automatic speaker verification (Dr. Kasorn, Dr. Pakinee, Dr. Sasiporn)
  • NICT's research on cross-domain spoof speech detection (Dr. Xugang Lu)

Registration

Welcome remarks

SIIT presentation

NICT presentation

Afternoon Session:
  • UCSY (Myanmar) presents on UCSYSpoof, a Myanmar language dataset for voice spoofing (Dr. Win Pa Pa, Hay Mar Soe Naing)
  • UBD (Brunei) discusses transfer learning for deep fake voice detection (Dr. Yassin, Navod Neranjan)
  • ITB (Indonesia) shares progress on an Indonesian language dataset for voice spoofing (Dr. Dessi Puji Lestari, Sarah Azka)

UCSY presentation

UBD presentation

ITB presentation

Discussion session

Day 2: ASEAN-IVO Technical Meeting

The second day highlights JAIST's research contributions, group discussion on existing works and the whole sessions:

  • Talk on Deepfake Speech Detection: Approaches from Acoustic Features (Prof. Masashi Unoki)
  • Discussions on the role of jitter and shimmer in fake audio detection (Dr. Li Kai)
  • Student presentations on audio classification and speech-pathological features for deepfake detection (Zaman, Anuwat)
  • Recent progress in ASVspoof challenges (Dr. Candy Mawalim, Aulia Adila)
  • Group discussion on current research issues in deepfake detection (All participants)

Prof. Unoki's talk

JAIST student presentation

Group discussion

Welcome dinner

Day 3: JAIST-ASEAN Deepfake Voice Detection Hub Symposium

On the third day, we held lab visits, open session and demonstration to all participants in JAIST.

  • Lab tours: Unoki laboratory and Okada laboratory
  • Board members' meetings on future collaborations and yearly research plans
  • Keynote sessions on Spoof Detection for Automatic Speaker Verification from Dr. Waree and Dr. Jessada
  • Open presentations and demonstrations from representatives of Thailand, Indonesia, Myanmar, Brunei, and Japan

Unoki-lab visit

Okada-lab visit

Keynote session: Dr. Waree

Keynote session: Dr. Jessada

Demonstration

Poster demonstration

The meetings were a resounding success, generating excitement and enthusiasm among all participants. The collaborative discussions and knowledge sharing fostered a vibrant atmosphere, laying the groundwork for future collaborations in spoof detection research.