How I improved my Korean listening skills

I self-study Korean. My routine is nothing special: a study session with a textbook or a video, handwritten notes as I go, and Anki reviews to keep what I covered from leaking away. It’s the routine half the self-study world runs on, and for reading and grammar it worked. Listening was the skill that stayed behind. I could read sentences I had no chance of catching by ear.

The standard advice is comprehensible input: listen to material you can mostly understand, and the gaps fill themselves in. I believe that advice. The problem was finding anything that was actually comprehensible to me.

Native podcasts were noise. I’d catch a couple words in a sentence and have nothing to hold on to — you can’t acquire language from a stream you’re 90% locked out of. The content I’d find on YouTube or the popular platforms for Korean language students was too comfortable, and not what I needed in that moment. In between sit the “intermediate learner” podcasts and playlists, which are calibrated to a fictional average learner — not to me. No published material was actually at the point I was at in my Korean journey (neither in vocab, grammar or topics that I wanted to comprehend better). How could it? More than that, I don’t even want to sum up all the time I’ve spent searching for content that was somewhat ok for me.

At some point I noticed where that information actually lives. My level isn’t “beginner” or “intermediate” — it’s the exact set of grammar points and vocabulary I’ve covered so far, nothing more. And the only place that set is written down is my own study notes. The notebook I was already filling after every session is a complete record of what I can, in principle, understand.

The pipeline

Having a technical background (data science & engineering), I decided to build a pipeline for myself. Gluing study notes, grammar books, OCR, LLMs, TTS, vector databases, etc together into a running profile of what I’ve studied. The result lands as an Anki deck: audio on the front, the Korean text and my related notes on the back. I import it and review it alongside the decks I already had. Apart from setting everything up as a side project, the whole thing now costs me about thirty seconds of effort per study session.

I want to be precise about what the audio feels like, because it’s the entire point. It is not easy listening. Written down, the texts would look almost trivial — it’s all grammar I’ve studied. By ear, they’re a small reach: I follow the thread, and once or twice per text I have to grab for something I only half-own. Somewhat understandable, by design. That narrow band — just past comfortable — is the band where listening actually improves, and it’s exactly the band no generic material can hit, because hitting it requires knowing what’s in my notes.

Did it work?

For me, yes, and faster than I expected, my listening improved noticeably within a couple of weeks, after being stuck for much longer than that. But I want to be honest about the size of that claim: the method has exactly one user, me. There’s no study, no cohort, no data beyond my own experience. That’s not evidence it transfers to anyone else. It’s the reason I turned the pipeline into an app — to find out.

So here is the offer, honestly described. The app is called up103. It’s run by one person; there are rough edges, and I read every piece of feedback myself. There are 50 Founding Learner spots, with a waitlist after that. Founding Learners get 6 months free, then 50% off for life — I’d rather state the terms plainly up front than surprise anyone later. What I’m hoping to learn is simple: whether the log-your-notes ritual that became automatic for me also works for people who aren’t me.

If you self-study Korean, already review with Anki, and listening is the skill that’s lagging, you can try it at app.up103.com.

— 에드, up103