Speak
Speaking-first AI tutor for spoken fluency
We publish a full Four-Mismatches fit assessment for the student-facing exam tools first. For this one, the facts below are what we have verified — run the Four Mismatches Test on it yourself for your own students.
Independence: Addestra builds learning products too (MathsTutor, TinkerWell). We never review or score our own, and no tool pays to be listed or to rank higher — how we assess fit.
What it does
What the AI actually does — A speech-recognition + LLM pipeline: users speak aloud and get real-time pronunciation and fluency feedback plus open conversation.
Why it was built — Replace expensive, hard-to-access human conversation practice with an always-available AI tutor focused on speaking.
Who built it
- Company
- Speakeasy Labs, Inc.
- Founders
- Connor Zwick · Andrew Hsu
- HQ / origin
- San Francisco, California, USA
- Founded
- 2016
- Model
- B2C
- Ownership
- Backed by the OpenAI Startup Fund
- Languages
- English (learners); strong in Korea/Japan)
- Pricing
- Subscription
We record durable facts only — no funding-raised or user-count figures. Unverified fields are marked [TBD] rather than guessed.
Where it sits
Speak is on the EdTech AI 100 map, in language learning, built in United States.
Compare it
Related & alternatives
- ELSA SpeakPronunciation and accent coaching for English
- LooraAI English-speaking coach for professionals
- PraktikaAI avatar tutors for spoken practice