Skip to content
TabBench

Speech to Text

Dictate and get a live transcript you can edit, copy, or download. Supports Hindi, Tamil, and more.

Uses your browser's speech recognition, which may process audio on your browser vendor's servers.

What the Speech to Text does

Speaking is roughly three times faster than typing, which makes dictation genuinely useful for first drafts, notes, and anything long. This turns speech into editable text live in the browser, with support for English, Hindi, Bengali, Tamil, Telugu, Marathi, Gujarati and more. One important caveat is stated up front: unlike the rest of this site, it is not fully private.

How to convert voice to text

  1. Choose your language. Accuracy drops sharply if the setting does not match what you are speaking.
  2. Press Start listening and allow microphone access when the browser asks.
  3. Speak normally. Words appear as you talk, with the in-progress phrase shown until it settles.
  4. Press Stop, then edit the transcript directly before copying or downloading it.

Uses your browser's speech recognition, which may process audio on your browser vendor's servers.

When to use it

Drafting faster than you can type

For a long email or a first draft, dictating and then editing is usually quicker than typing and editing, particularly on a phone.

Capturing thoughts while away from a keyboard

Notes taken by speaking are notes that actually get taken. Editing afterwards is far easier than reconstructing them from memory.

Typing in Indian languages without a keyboard layout

Dictating Hindi, Tamil, or Bengali avoids installing a keyboard layout or hunting for transliteration, which is where most people give up.

Good to know

  • Say punctuation aloud — 'comma', 'full stop', 'new paragraph' — as most engines recognise these commands.
  • A quiet room and a close microphone matter more than an expensive one.
  • Speak in complete phrases rather than word by word; the engine uses surrounding context to choose between similar-sounding words.
  • Firefox does not implement speech recognition. Use Chrome, Edge, or Safari.

Guides that use this tool

Frequently asked questions

Is my voice sent to a server?

In Chrome and Edge, yes — they use a cloud speech service. Safari transcribes on-device. This is a browser behaviour, not a site choice.

Is my voice sent to a server?

In Chrome and Edge, yes. Those browsers stream the captured audio to a cloud speech service to transcribe it, and neither this site nor any website using the standard API can change that. Safari performs recognition on-device. Because of this, dictating confidential material is not advisable, and this tool does not carry the client-side privacy badge the rest of the site uses.

Why does it stop listening on its own?

Browsers end a recognition session after a stretch of silence, and some cap the total length. Press Start again to continue — previously transcribed text is kept, so nothing is lost.

Which languages are supported?

Fourteen are offered here, including English in US, UK, and Indian variants, plus Hindi, Bengali, Tamil, Telugu, Marathi, Gujarati, Spanish, French, German, Arabic, and Japanese. Accuracy varies by language and accent, and the English variants matter — 'en-IN' recognises Indian English considerably better than 'en-US' does.

How accurate is it?

Good for clear speech in a quiet room — typically well above ninety percent for common languages. It degrades with background noise, strong accents, technical vocabulary, and proper nouns. Treat the output as a draft to edit rather than a finished transcript.

Does it work on a phone?

Yes, in Chrome on Android and Safari on iOS. Phone microphones are close to your mouth, which often makes accuracy better than on a laptop.

Related tools