> For the full page index, fetch https://echonova.ai/llms.txt

# Live transcription

Switch the Speech to Text page to Live, speak, and watch the text arrive after each pause. Learn what it costs, how long a session runs and where it is saved.

**Live** mode turns what you say into text as you say it. You speak, pause,
and the sentence appears. Use it for dictation, a meeting you want to follow,
or a quick check of how a dialect reads. For a recording you already have, use
[Transcribe a file](/docs/playground/transcribe-a-file).

## What you need

- A microphone, and a browser that can use it: Chrome, Edge or Safari.
- Credits. Live is charged per second of speech. See
  [Speech to Text limits and credits](/docs/playground/stt-limits-and-credits).

## Steps

### Step 1: Switch to Live

Open **Speech to Text** and press **Live** in the switch at the top of the
box. The status line reads **Ready to listen**. The box shows **Press Start
live and speak. The text arrives after each pause, in the dialect it was
spoken.** The button on the right reads **Start live**, with **Charged per
second of speech** under it.

You cannot switch while a take is being recorded in **File** mode. The page
says **Stop the recording first.**

### Step 2: Set the language and your terms

Pick **Language**, **العربية** or **English**, before you start. It is off
while a session runs. **Numbers as digits** and **Identify speakers** are for
files and do not appear in **Live**.

The session uses your terms as they are when you press **Start live**. A change
made while it runs applies to the next session. Press **Edit** in the
**Terminology** row to change them. See [Your terms](/docs/playground/your-terms).

### Step 3: Start and speak

Press **Start live**. The first time, the browser asks for permission to use
the microphone. The status line then follows the session:

| Status line                      | What it means                                                                          |
| -------------------------------- | -------------------------------------------------------------------------------------- |
| **Connecting…**                  | The page is opening the session.                                                       |
| **Listening**                    | Ready for you to speak.                                                                |
| **Hearing you**                  | You are speaking.                                                                      |
| **Transcribing…**                | Your last sentence is on its way.                                                      |
| **Reconnecting…**                | The connection dropped. The page is trying again. What you say now is not transcribed. |
| **Finishing the last sentence…** | You pressed **Stop**. The last words are arriving.                                     |

A timer beside the status line counts the session. Pause between sentences.
Each pause sends what you said, and the text appears as a new line with the
time it arrived. Words your terms corrected, and close matches, are
highlighted. To accept or dismiss a close match, open the saved transcript.

Press **Stop** to end the session. The words you already said still arrive.

If you open another page while a session runs, the page asks **Stop live
transcription?** Press **Stay** to keep going, or **Leave** to stop. The text
so far is saved. Closing or reloading the tab shows the browser's own warning.

<!-- screenshot: live-transcription/1 the page in Live mode while listening, with lines of text and the timer -->

### Step 4: Copy or download the text

Under the lines are three buttons: **Clear**, **Copy** and **TXT**. **Copy**
copies the text. **TXT** saves it as `EchoNova_Live_` followed by today's
date, for example `EchoNova_Live_2026-11-03.txt`. Both use your terms, so
they hold the corrected text. **Clear** empties the box and is off while a
session runs. It does not touch the copy that was saved to your transcripts.

## Cost and limits

- You are charged per second of speech. Long pauses are not sent and not
  charged. Each line is charged on its own, rounded to a whole credit, at least
  1.
- A session stops at 25 minutes. The page says **Live sessions stop at 25
  minutes. Start a new one to continue.**
- A session also stops after 5 minutes without speech. The page says **Live
  stopped after 5 minutes without speech. Start again when you are ready.**
- Only one live session can run at a time for your account.

## Where it is saved

When the session ends, the recording and its text are saved to your
transcripts. The line under the box reads **Saving…**, then **Saved to your
transcripts**, followed by the name and an **Open in Assets** link. The row in
the **Transcripts** table carries a **Live** tag. Open it like any other
transcript. See [Read a transcript](/docs/playground/reading-results).

If the page had to reconnect, the part before the drop is saved as its own
transcript.

If you said nothing, the page says **Nothing was said, so nothing was saved.**

## Troubleshooting

### Allow the microphone in your browser, then start again.

The browser blocked the microphone. Allow it in the site settings, then press
**Start live**.

### This browser cannot use the microphone. Use Chrome, Edge or Safari.

Open the page in one of those browsers.

### You are out of credits. Add credits to keep transcribing.

The balance ran out. Add credits from **Subscription & billing** in the user
menu. What arrived so far is kept.

### The connection dropped. What arrived so far is kept.

The page tried to reconnect and could not. Press **Start live** to begin a new
session.

### Live stopped after 5 minutes without speech. Start again when you are ready.

No speech was heard for 5 minutes. What arrived so far is saved. Press
**Start live** to go on.

### A live session is already running in another tab.

Stop the session in the other tab, then start again.

### Not saved. Copy the text before you leave this page.

The session ended but could not be saved. Press **Copy** or **TXT** before you
leave.

### We could not confirm the save. Check Assets in a minute.

The page did not hear back in time. The session may still have been saved.
Open **Assets** in a minute.

### Other messages

- **Your session expired. Sign in again.** Sign in, then start again.
- **The microphone stopped working.** Check the microphone, then start again.
- **Live transcription is not enabled yet.** Live is switched off for everyone
  for now. Use **Record** in **File** mode instead.
- **The transcription service is unavailable right now.** Try again in a few
  minutes.
- **Live transcription stopped. Try again.** The session ended for another
  reason. Start a new one.

## FAQ

### Why does the text wait until I pause?

The page sends each stretch of speech when you stop talking. A long run
without a pause is cut after about 8 seconds, so the text keeps coming.

### Can I use Live in Arabic and English?

Pick one language per session. Arabic covers every dialect, Fusha and mixed
Arabic and English.

### Is the audio kept?

Yes. The recording is saved with the transcript, so you can play it from the
**Transcripts** table. It holds only the speech: long pauses are left out, so
it is shorter than the session.

## Related

- [Transcribe a file](/docs/playground/transcribe-a-file): upload or record, then transcribe.
- [Your terms](/docs/playground/your-terms): spell names, products and acronyms your way.
- [Read a transcript](/docs/playground/reading-results): open the saved session.
- [Speech to Text limits and credits](/docs/playground/stt-limits-and-credits): the live rate and every limit.
