Curated sentence + native audio corpus for language learning
Problem
Language learners who use the mass-sentence method lost their reference tool: Glossika is enshittifying — the new app forces recording, removed listen-only mode, has no list of learned sentences, and leans into unverified AI-generated content. There is no source for accurate, human-curated sentence pairs with real native audio (not TTS).
Opportunity
A service offering curated bilingual sentence decks with authentic native speaker audio, spaced-repetition flow, listen-only mode for hands-free practice, and transparent human verification instead of LLM-generated content.
Market analysis
A real exodus is underway: the HN thread shows active Glossika refugees hunting for human-curated sentence pairs with genuine native audio, and no incumbent answers it — Clozemaster scales via mass sentences but leans on synthetic audio, and alternatives redirect to video- or chat-based learning. The blocker is content economics: recording and verifying thousands of sentences per language pair is slow, expensive, and at odds with solo-builder capacity.
Market · Intermediate/advanced mass-sentence learners (a dedicated but narrow subset of language learners); demand is event-driven — each Glossika misstep pushes more paying users to look for an exit.
Pricing · Comparables: Clozemaster Pro at $12.99/mo or ~$80/yr with a ~$199 lifetime tier; Lector from $5/mo managed. A premium human-verified corpus can plausibly charge Clozemaster-plus pricing, around $15-20/mo.
Pros
- + Documented demand spike from dissatisfied Glossika users actively seeking alternatives.
- + Human verification + real native audio is a defensible quality moat against AI-generated competitors.
- + Listen-only/hands-free mode is a cheap-to-build feature the incumbent just removed.
Cons
- − Native-audio recording and verification at scale is expensive per language pair and slow to expand.
- − Niche audience: mass-sentence method users are a small fraction of language learners.
- − Modern TTS keeps improving, steadily eroding the 'not TTS' differentiator.
- − Free stacks (Anki + community sentence decks) cover much of the need at zero cost.
Existing / similar tools
- → Clozemaster ↗
- → Lector ↗
- → TubeVocab ↗
Source
Hacker News (Ask HN)
The mistake would be competing with Glossika’s breadth. The winnable version is one language pair done obsessively well — curated CEFR-graded decks, verified native recordings, provenance shown per sentence — launched directly into the refugee community that is already asking for it. Pricing power comes from verification being visible: if every sentence links to its human checker, the product justifies a premium that AI-content competitors structurally cannot match. Expand to a second language pair only after the first is profitable.