Researcher. Open-source builder. App maker.

Eric Lam.

Publishing as Ho-Lam Chung

Speech, language, and things you can use.

I study how AI listens, speaks, and understands language. I lead OpenFormosa, an open AI initiative rooted in Taiwan, and build apps that bring ideas into everyday use.

Ph.D. student, National Taiwan UniversityPreviously a research intern at Microsoft Research.
01Read the papers 02Try the apps 03Build with the code

01 / Research

Listening. Speaking. Understanding.

My work connects speech, language models, and evaluation. These are selected papers, published under Ho-Lam Chung.

All papers on Google Scholar
  1. arXiv preprint · 2026Speech synthesis

    BlueMagpie-TTS: A Token-Efficient Tokenizer, Language Model, and TTS for Taiwanese-Accent Code-Switching Speech

  2. arXiv preprint · 2026Speech recognition

    Listen, Think, Transcribe: Continuous Latent Test-Time Scaling for ASR

  3. Findings of ACL · 2026Audio tokenization

    LLM-Codec: Neural Audio Codec Meets Language Model Objectives

  4. arXiv preprint · 2025Audio understanding

    TAU: A Benchmark for Cultural Sound Understanding Beyond Semantics

  5. SIGIR · 2025Deep research

    Researchy Questions: A Dataset of Multi-Perspective, Decompositional Questions for Deep Research

  6. arXiv preprint · 2025Speech recognition

    A Self-Refining Framework for Enhancing ASR Using TTS-Synthesized Data

  7. arXiv preprint · 2025Speech synthesis

    BreezyVoice: Adapting TTS for Taiwanese Mandarin with Enhanced Polyphone Disambiguation: Challenges and Insights

  8. Findings of ACL · 2024Audio evaluation

    Codec-SUPERB: An In-Depth Analysis of Sound Codec Models

02 / OpenFormosa

AI that understands where we come from.

I lead OpenFormosa. We build open models for Taiwan’s language, voices, and culture, from tokenization and language modeling to speech synthesis.

Visit OpenFormosa
Language model

Barbet

A compact 1B hybrid language model that provides a shared foundation for Taiwan-focused language and speech research.

Explore the model
Speech synthesis

BlueMagpie-TTS

Taiwanese-accent speech synthesis for local names, everyday language, and mixed Chinese and English text.

Listen to the demo
Tokenizer

PangolinTokenizer

A byte-level BPE tokenizer built for Taiwan-context text, shared by the OpenFormosa model family.

Get the tokenizer

03 / Independent apps

Small apps. Everyday possibilities.

Music, creative tools, and quieter digital spaces. 10 apps for iPhone, iPad, Mac, and Apple Watch.

See all apps on the App Store

04 / Open source

Ideas are better when they’re shared.

Language models, voice AI, research skills, Taiwan data, and developer tools. Explore the projects I build and share on GitHub.

Follow @voidful on GitHub

Keep in touch

Let’s build something useful.

Connect on LinkedIn