> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tryhamsa.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Live speech to text: see the text as you speak

> Run a live session that transcribes your microphone or a playing file as it goes, with speakers told apart, then copy, export or save it as a transcript.

> [!NOTE]
>
> Live speech to text is in beta and switched on for selected accounts.

A live session transcribes while you speak: text appears a moment after each pause, one line per phrase, with each voice given its own speaker. You can also play a file and watch it being transcribed as you hear it. When you stop, copy or export the text, or save the session as a regular transcript to edit it.

## Before you start

* A microphone the browser may use. The browser asks the first time; allow it. To transcribe a file instead, you don't need one.
* Credits left on the project owner's account. With none, the session doesn't start: "Your wallet is empty. Top up to use live transcription."
* The Hamsa Media tab open for the whole session. Nothing is kept if you leave the page without saving.

## Start a live session

**1. Open it**

In **Speech to text**, click **Record live** (marked **Live**). The **Live session** page opens.

![The three cards on the Speech to text page, with Record live marked Live as the third](https://mintcdn.com/hamsa-452fa79a/fdE-9_eEW1W6yh_s/images/media/speech-to-text/record-live-card-light.webp?fit=max\&auto=format\&n=fdE-9_eEW1W6yh_s\&q=85\&s=26cb0d2a71c4ea0323501ad6dc1f5db4)

**2. Pick and test the microphone**

Choose the **Microphone** and click **Test microphone**. Say something: the level moves and "Hearing you" shows when it picks you up.

**3. Start**

Click **Start recording** and speak. The status changes from **Connecting** to **Listening**, and **Hearing speech** while you talk.

![Start a live session: the microphone picker with its level meter and Hearing you, Start recording, and Play a file instead below](https://mintcdn.com/hamsa-452fa79a/fdE-9_eEW1W6yh_s/images/media/speech-to-text/live-start-light.webp?fit=max\&auto=format\&n=fdE-9_eEW1W6yh_s\&q=85\&s=1bbdea25e4abd5e89783dc8926f7bbed)

You don't choose a language: the session recognises Arabic and English by itself.

### Transcribe a file as it plays

Instead of **Start recording**, click **Play a file instead** and pick a WAV, MP3, MP4 or WEBM file up to 25 MB. You hear the file while it's transcribed, and pausing the file pauses the transcription. If the browser holds back the sound, the file plays muted; unmute it in the bar at the bottom.

## During the session

* Each phrase appears on its own line with its speaker, the time and, when there is enough audio, Male or Female. Text in grey is still settling and becomes final a moment later.
* **Speakers** lists each voice once it's heard. Rename one to rename every phrase by that speaker.
* **Edit phrase** corrects a line. "Search the transcript" finds text in the session. **Newest** jumps back to the latest line.
* **Pause** and **Resume**, or press Space. While paused, nothing is sent.
* **Stop** ends the recording and keeps the session on the page.
* You can switch microphone while recording; the session carries on with the new one.

![A live session in progress: phrases from Speaker 1 and Speaker 2 with their times, the newest still grey, Speakers listing Male and Female, and Hearing speech, Pause and Stop in the bar](https://mintcdn.com/hamsa-452fa79a/fdE-9_eEW1W6yh_s/images/media/speech-to-text/live-session-light.webp?fit=max\&auto=format\&n=fdE-9_eEW1W6yh_s\&q=85\&s=7c0569a49aa3257cefc7ca5c7c7a5f85)

If no sound arrives for a few seconds, the page warns that it hears nothing from the microphone. If the connection drops, you see "The live connection dropped. Resume to go on."; click **Resume**.

## After you stop

| Action | What it does |
| - | - |
| Play | Listen back to the session (Space plays and pauses) |
| **Copy** | Copies the text: each phrase with its speaker and time |
| **Export** | **Transcript (.txt)** or **Audio (.wav)** |
| **Continue recording** | Records more into the same session |
| **Save as transcript** | Adds the session to your transcripts (see below) |
| **Discard** | Deletes the text and the audio of this session |

![A stopped live session with its phrases, and the bar with play, Discard, Copy, Export, Continue recording and Save as transcript](https://mintcdn.com/hamsa-452fa79a/fdE-9_eEW1W6yh_s/images/media/speech-to-text/live-stopped-light.webp?fit=max\&auto=format\&n=fdE-9_eEW1W6yh_s\&q=85\&s=5632cbf9821e314a246da56298e89862)

## Save the session as a transcript

**1. Open the dialog**

Click **Save as transcript**.

**2. Name it**

Change the **Name** (it starts as "Live" and the date), up to 20 characters, and pick the **Language** spoken.

![The Save as a transcript dialog with the Name Live Oct 29, 16:40 and the Language, with Save](https://mintcdn.com/hamsa-452fa79a/fdE-9_eEW1W6yh_s/images/media/speech-to-text/live-save-dialog-light.webp?fit=max\&auto=format\&n=fdE-9_eEW1W6yh_s\&q=85\&s=a5558e515679be5e1bdc957e8227c834)

**3. Save**

Click **Save**. "Saved. The transcript is being prepared." The recording (or the file you played) is uploaded and transcribed again as a regular job, billed like an uploaded file.

The saved transcript appears in the list as **Pending**, then opens in the [transcript editor](https://docs.tryhamsa.com/media/speech-to-text/transcript-editor.md) like any other, with its own segments and speakers. Your live edits and speaker names don't carry over.

A session must be at least 30 seconds long to save: "Record at least 30 seconds to save it as a transcript."

## Good to know

* A live session belongs to the project it started in. Switching project discards it.
* Live text can differ from the saved transcript, which is transcribed again from the full recording.
* If one phrase fails, the session goes on: "Speech to text failed on that phrase. Keep talking, the session is still on."

## Troubleshooting

| Message | What to do |
| - | - |
| "The browser blocked the microphone. Allow it in the site settings." | Allow the microphone for this site in the browser's settings, then try again |
| "No microphone found. Plug one in or pick another." | Connect a microphone or choose another in **Microphone** |
| "No sound from … Check it isn't muted, or pick another microphone." | Unmute the microphone or pick another |
| "This file is over 25 MB, the live transcription limit." | Use a shorter file, or [upload it](https://docs.tryhamsa.com/media/speech-to-text/transcribe-files.md) as a regular transcript |
| "You're not a member of this project, so live transcription isn't available here." | Ask an Owner or Admin to invite you |
| "Live transcription couldn't connect. Try again." | Check your connection and start again |

## Related

> [card] [Transcribe files and links](https://docs.tryhamsa.com/media/speech-to-text/transcribe-files.md)
>
> Upload, paste a YouTube link or record.

> [card] [Transcript editor](https://docs.tryhamsa.com/media/speech-to-text/transcript-editor.md)
>
> Edit a saved session.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.