Kugel Ear 1

Kugel Ear 1 is KugelAudio's streaming speech-to-text model for voice agents that flags words it is unsure of with alternative spellings, waits out mid-sentence pauses before ending a turn, and costs €0.48 per hour.

Why it matters

For businesses running phone bots, it targets the usual failure points (misheard names, numbers and callers getting cut off) so the bot can ask once instead of making callers spell things repeatedly.

Highlights

• Sends the LLM a confidence flag and alternative spellings for words it may have misheard; KugelAudio says 86% of wrong words were flagged on its synthetic German call test.unverified
• Built-in end-of-turn detection waits out pauses inside a sentence; KugelAudio reports the fewest false cut-offs on LiveKit's open end-of-turn benchmark.
• KugelAudio reports 93.4% words right on AppTek's public call-center set and a 72 ms median time to final transcript.unverified
• One WebSocket stream: audio in, final transcript when the turn ends, then the word alternatives.
• Covers nine European languages, set per stream or auto-detected per turn.unverified

Taken from the project's own README, docs or launch page. Claims about speed or accuracy are the maker's unless stated otherwise.

Worth knowing: All benchmark numbers are KugelAudio's own runs; the names-and-addresses test uses synthetic German audio KugelAudio built and has not published.

Quick facts

Links

• Official site: https://www.kugelaudio.com/en/kugel-ear-1
• Release notes: https://docs.kugelaudio.com/changelog
• Docs: https://docs.kugelaudio.com/

Related ranking

See how established tools compare in Best Voice & Text-to-Speech. New releases aren't scored until we review them against the rubric.

What we couldn't verify

• highlights[0] (vendor benchmark)
• highlights[2] (vendor benchmark)
• highlights[4] (which nine languages)

Checked 2026-10-22 from the model's own pages. Prices and star counts change, so confirm on the official site. Listing here is not a ranking or an endorsement.