大声读 · ReadAloud

What it does

Speech recognition runs locally, and Jev sees only the transcript, so it cannot judge pronunciation, stress, or accent. Local ASR hardware is required.

How you can use it

You can build a read-aloud practice tool that catches reading mistakes as someone speaks. First, collect sample English passages for your learners to practice reading aloud. A developer can then set up a speech model on your computer to turn live spoken words into written text without sending audio online.

Your developer connects the text to ZenMux using an account access key to check whether words were read correctly and score the passage meaning. Because the judging service only sees written text, it cannot grade speech sounds like pronunciation, vocal stress, or accents.