Voice-to-writing glossary

You talk. Software listens. It writes down what you said. Some speech-to-text runs in the cloud on large servers. Some runs on your device. Some, like the browser Web Speech API that Zahvox uses today, is built into your browser and handled by the browser vendor.

You speak. It writes. The whole point is to skip the keyboard — either because you can't type easily, don't want to, or just think faster than you type.

It's dictation, but tied to one text field. You press a mic button, talk, and the words show up where the cursor is.

You have a recording. You want the words in text form. That's transcription.

It's a small door in your browser. Zahvox knocks on the door and says 'please listen to the mic and tell me what you hear.' The browser handles the actual listening.

Take what the recognizer wrote. Compare it to what was actually said. Count the mistakes. Divide by total words. That's WER.

You tap record, talk for 30 seconds, tap stop. The recording is the note.

Open a tab. Click a mic. Talk. See text. All without installing anything.

Talk. Get text. Fix a couple of things. Send it. That's voice-to-writing.

The recognizer gives you rough text. Cleanup makes it look like something you'd actually send.

It's a text box that your voice fills in. You can change any word, any moment, without starting over.

Copy: your text goes to the clipboard, ready to paste anywhere. Download: your text becomes a small file you can keep or attach.

The whole path from 'I have a thought' to 'that thought is sent, published, or shared.'

It's not the final version. It's a starting point you can improve.

The message you send back to a client. Every day. Sometimes ten times a day.

The write-up that comes out of a meeting so everyone knows what was decided and what to do next.

You speak messy. The AI tidies. Your meaning stays. Your voice stays.

You give it an instruction; it writes for you.

You pick a mode; the output looks like that kind of writing.

Same content, different feel. 'Friendly' sounds like a coffee chat. 'Formal' sounds like a board update.

A shape for your writing. You provide the raw thoughts by voice; the template arranges them.

What you tell the AI to do.

The short version.

Taking what's already written and saying it differently.

A small program that lives inside your browser and gives it new abilities.

Instead of copy-paste, the text just appears where you're typing.

The browser asks: 'let this site hear you?' You say yes or no. That's the permission.

A small drawer in your browser where a website can keep notes for itself.

A button on a website that opens your phone or computer's normal share menu.

The list of things the browser asks a website to prove it should be allowed to do.

A log of everything you've dictated, so you can go back and find it later.

A Zahvox app that lives on your computer, so voice-to-writing works everywhere — not just in a browser tab.

A Zahvox app for your phone that fits naturally into iOS and Android workflows.

Small pieces of Zahvox that live inside other apps.

A shared version of Zahvox for a team — same templates, same tone, same account.

The list of what a product has, what it's building next, and what it might build after.