Turns native-language audio into flashcards and looping shadowing practice.
Last verified · 2026-06-26 · by Moe Ameen
LingoChunk is a browser tool that turns a piece of spoken audio into language-study material. It was built by a solo developer (Hacker News user "alder") to help with their own German and Greek studies, and shared as a "Show HN" on 25 June 2026. It started as a way to make Anki flashcards from native-language audio and grew into a small workbench for listening, mining vocabulary, and practicing pronunciation.
The core pipeline is transcription with word-level timestamps. LingoChunk transcribes the audio, extracts the words, finds their base forms (lemmas), and groups example sentences by lemma — so instead of a flat transcript you get vocabulary organized by dictionary form, each with real spoken examples. Because every word is time-stamped, the tool can also do things a plain transcript can't: it generates cloze-style flashcards where the target word is blurred on the front and revealed on the back, and it can play the example sentence with the word "beeped" so you test recall against the audio, not just the text. Cards live in a built-in spaced-repetition (SRS) system and can also export to Anki.
A second mode is shadowing. You click the first and last word of any passage in the transcript and LingoChunk loops that segment with a configurable gap and repeat count, so you can repeat after a native speaker to drill rhythm and pronunciation. The developer notes they also use it as a plain pronunciation player to improve their own English.
There is also an AI helper: you select an audio fragment and send a predefined prompt — "explain grammar" or "explain nuances of meaning" — to get a short explanation of the selected line. The developer describes those prompts as still experimental. LingoChunk supports 15 input languages (Japanese and Chinese are recent, experimental additions) and more than 30 output (explanation) languages. The transcription back-end (Soniox) covers many more languages than the tested set, so coverage may widen over time. Treat specifics as a moving snapshot of an early, solo-built project; pricing was not publicly detailed at launch beyond a free "try" page seeded with public-domain audio.
LingoChunk is a private study tool — it makes *you* better at a language. But the people who use a tool like this most are often language-teaching creators, one of the largest evergreen niches on TikTok, Reels, Shorts, and YouTube, and the study artifacts LingoChunk produces are excellent raw material for lessons. That is where Kompozy comes in: it turns a study insight into a finished, on-brand lesson published across platforms. A single "explain grammar" answer becomes a Carousel that walks through the rule slide by slide. A lemma group — five real ways natives use one word — becomes a Listicle Video over a clean portrait clip, a Quote Graphic set, or a text post in your teaching voice. A shadowing passage becomes a Persona Short where your avatar models the phrase and Kompozy burns in the captions.
The leverage is the fan-out and the publish. One audio source you mined in LingoChunk can seed a week of cross-platform teaching content in Kompozy — short video for the feeds, a carousel for the save, a thread for X, a "phrase of the week" newsletter, and a blog post for search — all written in one consistent voice through your Persona Brief, then scheduled and published to all nine platforms from a single queue. LingoChunk helps you understand the language; Kompozy is how you teach it to an audience at the volume the algorithms reward.
LingoChunk is a browser tool that turns native-language audio into study material: it transcribes the audio with word-level timestamps, mines vocabulary grouped by base form (lemma), makes cloze flashcards you can export to Anki, and loops passages for shadowing practice. It was built by a solo developer and shared as a Show HN in June 2026.
You click the first and last word of a passage in the transcript, and LingoChunk loops that exact audio segment with a configurable gap and repeat count. You repeat after the native speaker to drill pronunciation and rhythm. The developer also uses it as a plain pronunciation player for their own English practice.
Yes. It generates cloze-style cards (the target word blurred on the front, revealed on the back) in a built-in spaced-repetition system, and those cards can also export to Anki. One detail: the audio "beep" over the target word works in the built-in player but is not currently part of the Anki export.
At launch it supported 15 input languages, with Japanese and Chinese as recent experimental additions, and more than 30 output languages for the explanations. The underlying transcription back-end covers many more languages than were tested, so coverage may grow. Check the site for the current list.
LingoChunk is for private study, not publishing. To teach an audience, take what you mined — a grammar explanation, a vocabulary set, a pronunciation passage — into Kompozy, which generates a carousel, a short video, a thread, a newsletter, and a blog from that one source in your voice, then schedules and publishes them across nine platforms.