Open source general-purpose speech recognition model that also handles speech translation and language identification, used via Python or command line.
check_circle Free plan available
Made for
Developers who want to run speech recognition themselves
Free use
Code and model weights are released under the MIT License.
How to choose
No tool is best for everyone. Try two or three on one real task of yours, then use this checklist.
Does it support your language and accent?Test with a real recording. Accuracy can vary a lot between languages and accents.
Where does your audio go?Recordings can contain private details. Read how the tool stores and uses uploads.
Can you use the result commercially?Read the licence for voices and music, especially if you plan to publish or sell.
Is cloning a voice allowed?Only clone a voice with the owner's clear permission. Check the tool's rules.
What is free?Check the free option, and whether exports are limited or marked.
Can you fix mistakes easily?For transcripts, look for an editor that plays the audio next to the text so you can correct errors.
Frequently asked questions
Are AI transcripts accurate?
They are often good but not perfect, especially with noise, accents and names. Proofread before you share.
Can I use AI music in my videos?
Read the tool's licence for your plan. Rules differ and can change.
Is it legal to copy someone's voice?
Not without permission. Voice cloning of a real person can break the law and platform rules in many places.