About Me: Yi-Chang Chen

Posted on June 06, 2026 in Life. View: 7,408

portrait

YC / Yi-Chang Chen / 陳宜昌

YC is the CEO & Founder of Noopher AI — a company dedicated to solving real-world problems backed by deep AI research. He has over 8 years of R&D experience in machine learning — spanning multiple industries (semiconductor, finance, and marketing) and roles (backend engineer, ML engineer, and AI researcher) — with a focus on language, visual, and speech AI. With 5 years of leadership experience leading R&D teams of up to 5 researchers, he has successfully brought projects from research to production, pioneering multiple Taiwan-first open-source AI systems — Breeze (LLM), BreezyVoice (TTS with voice cloning), and BreezeASR-Taigi (Taigi speech recognition). He has published 8 AI papers at top-tier conferences including ICLR (one of the top venues in machine learning), ACL, and INTERSPEECH, accumulating 100+ academic citations.

OPEN-SOURCE PROJECTS
📌 Breeze — Taiwan's first open-source Traditional Chinese LLM (103k+ downloads)
📌 Breeze2 — Next-gen Traditional Chinese LLM with vision understanding, OCR, and function-calling capabilities for agentic use; available in 3B and 8B (77k+ downloads)
📌 BreezyVoice — Taiwan's first open-source TTS with voice-cloning capability
📌 BreezeASR-Taigi — SoTA Taiwanese Hokkien (Taigi) ASR model (9k+ downloads)
📌 g2pW — SoTA Mandarin grapheme-to-phoneme model (adopted in PaddleSpeech & pypinyin)

EXPERIENCE
📌 Noopher AI, CEO & Founder | May 2026 - Present
📌 MediaTek Research, Senior ML Researcher / Senior Technology Manager | Mar. 2023 - May 2026
📌 E.Sun Commercial Bank, ML Engineer / Lead ML Researcher | Mar. 2020 - Mar. 2023
📌 Bridgewell, ML Engineer | Mar. 2018 - Mar. 2020
📌 Ur Warm Pet Cafe, Resident Singer | Mar. 2017 - Mar. 2018
📌 TSMC, R&D Engineer | Nov. 2015 - Mar. 2018

PUBLICATIONS (5 SELECTED FROM 8)
📌 “TASTE: Text-Aligned Speech Tokenization and Embedding for Spoken Language Modeling,” in ICLR, 2026. [Paper][Code]
📌 “Let's Fuse Step by Step: A Generative Fusion Decoding Algorithm with LLMs for Robust and Instruction-Aware ASR and OCR,” in Findings of ACL, 2025. [Paper][Code]
📌 “Enhancing Function-Calling Capabilities in LLMs: Strategies for Prompt Formats, Data Integration, and Multilingual Translation,” in NAACL, 2025. [Paper]
📌 “g2pW: A Conditional Weighted Softmax BERT for Polyphone Disambiguation in Mandarin,” in INTERSPEECH, 2022. [Paper][Code]
📌 “Integrated Semantic and Phonetic Post-correction for Chinese Speech Recognition,” in ROCLING, 2021. (Best Paper Award) [Paper][Code]