Hi Product Hunt! I'm the developer of Ether Echo. I built it because of a frustration: I spent months on a language app and could recognize hundreds of words on a screen, but I froze the moment someone spoke to me. Tapping is not listening, and it is definitely not speaking. So Ether Echo does only those two things. You hear whole sentences said by a voice, and every sentence introduces exactly one word you don't know yet. Each word comes back right before you would forget it, scheduled with FSRS, the algorithm recent Anki versions use. Once you know about thirty words, a real-time voice tutor talks with you, and it stays inside the vocabulary you actually have, so you can hold a first conversation on day two instead of month six. A few honest details: - Seven languages: English, French, Spanish, German, Japanese, Mandarin, Russian. - Non-Latin scripts are shown as romaji, pinyin with tones, and stressed transliteration, so you always read something you can pronounce. You will not learn to write here, and that is on purpose. - Learning and training are free, for good. The tutor is the paid part, because a real voice and a real model in real time cost real money. Thirty minutes free at sign-up, no card. - Built by two people in Montreal. Stack, for the curious: TypeScript monorepo, Postgres, a Python voice pipeline (Pipecat, LiveKit, Cartesia), Expo for mobile, Next.js for web. It's a first version and it launched four days ago, so I'm looking for testers more than anything. Tell me what gets in the way, what's missing, what you would have done differently. I'll answer every comment.

Ether Echo

Learn a language by ear, then talk to a voice tutor

Learn a language by ear, then talk to a voice tutor