Mei Lin Chen

Speech Recognition Engineer, Live Subtitles — based in Singapore

Mei Lin owns the speech-recognition pipeline behind Live Subtitles. She is responsible for the parts of the system that decide whether captions feel "almost-right" or genuinely accurate, especially on the messy speech most ASR demos avoid: overlapping speakers, slang, and strong regional accents.

Background

Mei Lin holds an MSc in computational linguistics from the National University of Singapore and previously worked on call-centre speech analytics for a regional telecom and on closed-caption pipelines for a Hong Kong broadcaster. She has written and shipped acoustic-model fine-tuning code for Mandarin, Cantonese, Singlish, and English-as-an-L2 speakers, which is why Live Subtitles handles non-textbook speech as a baseline expectation, not as a stretch goal.

What Mei Lin writes about

Why generic ASR fails on conversational audio, and what tooling changes the outcome.
How non-native learners can use subtitles to decode slang, idiom, and fast-speech reductions.
Honest assessments of when subtitles help native speakers (accessibility, focus, learning a new accent) and when they get in the way.
Practical mistakes learners make when judging an article subtitle as "wrong" without checking the spoken context.

Editorial standards

Mei Lin's articles cite measurable behaviour (word error rate by audio condition, retention rate across spaced sessions) rather than vague qualitative claims. Where the product has a real limitation, she states it explicitly — this is the easiest way to keep technical writing trustworthy.