ZYSNOR, smart solutions, real impact
Speech to Text

Writing guide

How to use speech to text

By , Founder of ZYSNOR · Updated · Checked · 7 min read

In short

Choose your language, press Start listening, allow the microphone and speak; the words appear as you talk and you can edit them. In English you can say comma, period, question mark, new line or new paragraph. By default browsers such as Chrome send the audio to an online recognition service, so it needs the internet and you should not dictate secrets.

Jump to a section

Dictation is faster than typing for most people, and it helps when typing is hard. Browser dictation is free, but it has one thing you should know first: where the audio goes.

Where your voice goes

By default, browser speech recognition uses an online engine: your audio is sent to a web service and the text comes back, which is also why it does not work offline. Chrome sends it to Google's servers. Newer browsers can run recognition on the device when a page asks for it. ZYSNOR does not record your audio, and your text is not sent to ZYSNOR: the auto-save copy stays in your own browser.

Four steps from microphone to recognition service to text to your edits
Where your voice goes.

Languages

The tool offers English (US and UK), Arabic, Hindi, Urdu, Bengali, Mandarin Chinese and Persian. Whether each works depends on your browser; Chrome and Edge support the most.

Punctuation by voice

In English, say the mark: comma, period or full stop, question mark, exclamation mark, colon, new line, new paragraph. The tool replaces the words with the marks. Other languages get the text as recognised; add punctuation afterwards.

A table of English voice commands and the punctuation they produce
English voice commands.

Worked example 1: a quick email

Say "hi Sara comma thanks for the file period new paragraph I will reply tomorrow period". You get "hi Sara, thanks for the file. (blank line) I will reply tomorrow." Capitalise the first word by hand.

Worked example 2: notes in Urdu

Choose Urdu and speak at a steady pace. Voice commands are English only, so add punctuation after you stop.

Worked example 3: the microphone is blocked

If the tool says the microphone is blocked, click the lock icon next to the address, allow the microphone for the site and reload.

Getting better accuracy

  • Use a headset or sit close to the microphone.
  • Cut background noise.
  • Speak in full sentences at a steady speed.
  • Check names, numbers and technical words; they are misheard most often.

What the ZYSNOR Speech to Text does

  • Eight languages, live words, editable transcript.
  • English voice commands for punctuation and new lines.
  • No recording by ZYSNOR; the browser's service may process audio online.

Practice problems

Problem 1. Can you dictate offline in Chrome?

Show the worked answer

Usually not. Recognition runs on an online service by default.

Problem 2. What do you say to start a new paragraph?

Show the worked answer

"new paragraph" (English, with voice commands on).

Ready to try it? The free tool works in your browser, and the button in the page header takes you straight to it.

Common questions

Is my voice recorded?

ZYSNOR does not record it. Your browser's speech service, for example Google's in Chrome, may process the audio online.

Which languages work?

English (US, UK), Arabic, Hindi, Urdu, Bengali, Mandarin Chinese and Persian, where your browser supports them.

Can I add punctuation by voice?

Yes in English: say comma, period, question mark, exclamation mark, colon, new line or new paragraph.

Why does it not work in my browser?

Not every browser supports speech recognition. Try Chrome or Edge.

Does it work offline?

Usually not, because recognition runs on an online service by default.

Can I edit the text?

Yes. The transcript box is editable.

Sources and how we researched this

Sources (3)

How we researched this

Behaviour was read from the tool's code (continuous recognition, interim results, eight languages, English voice commands, microphone error messages); dictation could not be tested in our headless browser, so accuracy was not measured. MDN states that by default speech recognition uses a server-based engine and the audio is sent to a web service, so it does not work offline, and describes an optional on-device mode some browsers now offer. Mozilla's wiki and a Twilio article say Chrome sends audio to Google's servers. Spotted something that does not match your experience? Tell us through the Contact page and we will check it and update this guide.

Related guides

Rules differ between schools and change over time. Always check your own school or university handbook.