About Me

Hello, I'm Punpun! I am currently an HAI visiting scholar at the Stanford NLP Group, working on mulitmodal (audio/speech) LLM research with Diyi Yang.

I am also a founding team member of Typhoon, an AI lab supported by SCB 10X to advance Thai AI models (e.g., LLM, ASR, TTS), where I currently focus on audio/speech models.

I completed my PhD in May 2024 from the University of Cambridge, supervised by Mark Gales on summarization and factual consistency of NLG. Previously, I had a research internship at Amazon, and I hold a B.A. and an M.Eng. from the University of Cambridge.

Recent Projects

My recent projects cover the following themes:

• Multimodal, Audio, Speech: AudioJudge, TalkArena, Low-Resource AudioLLM, ASR Error Correction with LLM
• NLG Evaluation, Factuality, AI Safety: SelfCheckGPT, CrossCheckGPT, MQAG, LLM Comparative Assessment, SkillAggregation, Unlearning vs. Obfuscation
• Typhoon Project (opentyphoon.ai): Typhoon-TTS, Typhoon2-Audio, Typhoon2-LLM, Typhoon-Audio, Typhoon-LLM, Thai-HELM

News

• [2025/08] EMNLP 2025: Unlearning vs. Obfuscation (main); Prior Prompt for RFT (main)
• [2025/07] Start at Stanford NLP as a visiting scholar
• [2025/05] Interspeech 2025: Typhoon-Audio is accepted
• [2025/05] ACL 2025: SkillAggregation (main), TalkArena (main); xPoT (findings)
• [2025/02] We released Typhoon Reasoning: T1 (long-though SFT) & R1 (Merging)
• [2024/12] We released 🌪️ Typhoon 2 || Text Models & Multimodal Models
• [2024/09] EMNLP 2024: 2*main + 1*findings as a co-author
• [2024/09] We released ThaiExam in HELM (w/ Stanford CRFM)
• [2024/08] Speak at AI CoP Seminar @ the U.S. SEC on LLM Hallucination
• [2024/08] Speak at 3rd SEA Language Summit on Multimodal LLM
• [2024/08] We released 🌪️ Typhoon Multimodal: Audio Model & Vision Model
• [2024/07] Speak at the Stock Exchange of Thailand (SET) on Thai LLM

Selected Publications

A complete list of publications is available on Google Scholar

• Enhancing Low-Resource Language and Instruction Following Capabilities of Audio Language Models
  Potsawee Manakul, Guangzhi Sun, Warit Sirichotedumrong, Kasima Tharnpipitchai, Kunat Pipatanakul
  Interspeech 2025 [project page]
• SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models
  Potsawee Manakul, Adian Liusie, Mark Gales
  EMNLP 2023 [project page]
  → Featured in the Washington Post about AI Hallucination in May 2023
  → Featured in deeplearning.ai short course on LLM safety
  → Featured in reading list of Stanford CS 329T: Trustworthy Machine Learning (Fall 2023)
  → Invited talk at ML Collective - Deep Learning: Classics and Trends (DLCT) in August 2023
  → Invited talk at at AI CoP Seminar @ the U.S. SEC on LLM Hallucination in August 2024
  → Featured in an analysis by Daniel Huynh (from Mithril Security) on SelfCheckGPT
• MQAG: Multiple-choice Question Answering and Generation for Assessing Information Consistency in Summarization
  Potsawee Manakul, Adian Liusie, Mark Gales
  AACL 2023 [code]
  → Area Chair Paper Award (Generation and Summarization)
• Long-Span Summarization via Local Attention and Content Selection
  Potsawee Manakul, Mark Gales
  ACL 2021 [code]
  → System winning 1st place in the Spotify Podcast Summarisation task 2020
• Abstractive Spoken Document Summarization using Hierarchical Model with Multi-stage Attention Diversity Optimization
  Potsawee Manakul, Mark J. F. Gales, Linlin Wang
  Interspeech 2020 [code]
  → Best Student Paper Award Finalist
  

---

Honours and Awards

• AACL 2023 Area Chair Paper Award
• InterSpeech 2020 Best Student Paper Finalist
• Cambridge International & St John's College Scholarship (2019)
• BP 1st Year Prize: top 1% in 1st-year Cambridge Engineering (2016)
• Thai Government Scholarship (2015)
• Silver Medalist at IPhO 2014; Gold Medalist at APhO 2014

Service

• Area Chair: ACL 2025
• ACL Rolling Review: 2024 (Feb, Apr, Jun, Oct, Dec), 2025 (Jul)
• NeurIPS: 2023, 2024, 2025
• ICLR: 2024, 2025
• ICML: 2023, 2024, 2025
• AAAI: 2025
• Interspeech: 2023, 2024, 2025
• Conference On Language Modeling (COLM): 2024, 2025
• IEEE Transactions on Audio, Speech, and Language Processing: 2022, 2023, 2024, 2025

---

Contact

• Email: potsawee@stanford.edu

---