About Me

Hello! I'm Arthur Chan — an independent AI advisor, lead researcher, and senior speech scientist.

For the past 25 years, I’ve worked on speech recognition–related problems — now more broadly known as Conversational AI.
My past roles include Staff Scientist at BBN, Speech Scientist at SpeechWorks (acquired by Nuance), Core Technology Scientist at ScanScout (acquired by Tremor Video), and Principal Speech Architect at Voci (acquired by Medallia).

At CMU

I was one of the maintainers of the CMU Sphinx project. I primarily worked on Sphinx 3.X, but later contributed to SphinxTrain as well.

At Voci / Medallia

As Principal Speech Architect, I specialized in maintaining and enhancing ASR-related software. I also worked on machine learning tasks including emotion classification, gender detection, and keyword spotting — balancing hands-on development with team leadership.

After Voci was acquired by Medallia, I took on a leadership role, managing a team of three (two PhDs and one MS). We were responsible for delivering language packs for the Voci V7 ASR system and developing version V9.

Independent AI Advisory

I now work independently as an AI advisor, focusing on Conversational AI — including ASR, large language models (LLMs), and text-to-speech (TTS).

Feel free to connect with me on LinkedIn for further inquiries. For consultation calls, you may schedule a zoom meeting with me on calendly.

Others

With Waikit Lau, I have curated and manage the Facebook AIDL Group and its LinkedIn version.

---

This page revives my old blog — The Grand Janitor Blog — where I shared thoughts on speech recognition, machine learning, and other tech topics.