Every language app I tried gave me one answer and moved on. That works until you're composing a work email in French and you genuinely don't know whether you're being too cold or too familiar — and there's no signal in a single translated string.
The sharper problem is audio. Written French doesn't show you vowel length or liaison patterns. Written Japanese doesn't show pitch accent. You can memorize a sentence perfectly and still hear a native speaker say it in a way that doesn't match what you drilled. Tone contour and vowel duration are what make a phrase stick — they give your memory something physical to anchor to, not just a character string.
I built konid to fix both of those things at once. For any phrase you want to say, it returns 3 options ordered casual to formal, with the register explained and a nuance comparison between them. Then it plays audio pronunciation directly through your speakers using node-edge-tts — no external API key, no extra setup.
That pairing matters: you read the register note ("option 2 is appropriate for colleagues you're friendly with but not close to"), then you immediately hear both prosody and cadence. The explanation gives you context; the audio gives you the muscle memory cue.
It runs as an MCP server so it drops into Claude Code, Cursor, VS Code Copilot, Windsurf, Zed, JetBrains, or Claude Cowork without switching context:
claude mcp add konid-ai -- npx -y konid-ai
Also available as a ChatGPT app via Developer mode at https://konid.fly.dev/mcp.
13+ languages including Mandarin, Japanese, Korean, Spanish, French, German, Portuguese, Italian, Russian, Arabic, and Hindi. MIT licensed.
github.com/robertnowell/konid-language-learning