ElevenLabs Speech to Text (2026): Transcribe Audio, Edit Text & Export Subtitles
ElevenLabs Speech to Text turns an audio or video recording into an editable transcript. You can then check the wording, label speakers and export text or subtitles. The useful part is not skipping the review. It is starting with a draft instead of a blank page.
Mike Says earns commission if you use our ElevenLabs affiliate link and make a qualifying purchase, at no extra cost to you.
Affiliate link · I may earn commission from a qualifying purchase at no extra cost to you.
Concept artwork, not a screenshot of ElevenLabs.
Published 8 October 2026. Based on official documentation checked today. This is a source-based guide, not a claim of hands-on transcription testing.
Is it worth trying?
Yes, if you regularly turn recordings into notes, written content or captions. Start with a short, non-sensitive clip. Check the result against the audio before deciding how much editing a longer recording will need.
If you only need one quick transcript, try the free allowance first. If your video editor already gives you usable captions, another subscription might not add much.
What this guide covers
This is the browser workflow for an existing recording. It does not cover an API integration or live microphone transcription. ElevenLabs lists Scribe v2 as its default model for uploaded recordings in the web interface; Scribe v2 Realtime is a separate live-transcription option.
A sensible first project is a short voice note you recorded yourself. Aim for clear speech and one speaker, then move to interviews or noisier files once you know what needs checking.
Before you upload
- Keep your original recording so you can compare the transcript with it.
- Only upload recordings you have permission to use. Follow your organisation's policy for client or internal material.
- Do not start with private client, medical or financial audio just to test the tool.
- Check your account's remaining credits and the upload's cost before committing a long recording.
- Write down the correct spelling of names, brands and specialist terms mentioned in the clip.
1. Upload your recording
Open Speech to Text in the ElevenLabs dashboard, choose Transcribe files and add your audio or video. Set the primary language if you know it, or use Detect. The web upload documentation lists a 3 GB file limit.
Optional audio-event tags can mark sounds such as laughter. Leave them off if your first goal is clean spoken text. Check the settings, then select Upload files to submit the job. When it is ready, open the file's name.
Start small: a two-minute clip is easier to review than an hour-long interview. Keep a note of the credits used and the editing time. That gives you a useful basis for planning your next upload, without relying on someone else's accuracy claim.
Affiliate link · I may earn commission from a qualifying purchase at no extra cost to you.
2. Check the words and speakers
Click a word to play the audio from that point. Listen while reading, rather than checking the text on its own. Edit incorrect wording directly in the transcript.
Give extra attention to the parts that change meaning:
| Check | Why it matters |
|---|---|
| Names and product names | A plausible spelling can still name the wrong person or tool. |
| Numbers, dates and prices | A single digit can change an instruction or quote. |
| Negatives such as “not” | A missed word can reverse the speaker's meaning. |
| Speaker changes | The right sentence under the wrong name is still incorrect. |
| Overlapping or quiet speech | Replaying a difficult section is better than filling it in from context. |
Rename generic speaker labels so the transcript is useful to someone else. You can also reassign a segment to another speaker. Do not delete a speaker just to change their name: the editor documentation says deletion also removes that speaker's associated segments.
For example, in a fictional interview with “Speaker 1” and “Speaker 2”, identify which voice belongs to each person before renaming either label. Then spot-check later sections, especially where both people speak at once.
If the recording is unclear, mark the uncertainty in your working copy and ask the speaker to confirm it. Do not turn a guess into an attributed quote.
3. Check timing and export
For captions, check segment timing as well as wording. The editor lets you adjust segment boundaries on the timeline or enter start and end times. After text changes, Align words recalculates word timestamps. Use Export when the transcript is ready.
| Format | Useful for |
|---|---|
| Plain text | A readable transcript for notes or a writing draft. |
| HTML | A structured text export for a browser-based workflow. |
| JSON | Structured data for a technical workflow. |
| SRT or VTT | Subtitle files for a compatible player or editing tool. |
Import the subtitle file into your intended editor or player and watch the finished sequence. Look for late captions, awkward breaks and lines that stay on screen too briefly. An exported subtitle file is not the same as captions permanently added to the video.
What does ElevenLabs Speech to Text cost?
The current Free plan is $0 and includes Speech to Text with 10,000 monthly credits shared across products. Compare all ElevenLabs plans and credits. That credit count is not a promise of a fixed number of transcription minutes.
Check the Speech to Text allowance and rate shown for your account and workflow. The pricing pages show separate website and API figures, so do not apply an API hourly price to a browser upload. Other ElevenLabs usage can also reduce your remaining shared credits.
If you need a paid plan, compare its transcription allowance with your actual workload. Check the renewal amount, taxes and any extra-usage options before subscribing. There is no reason to pay just to finish this first test if your free balance covers it.
A simple quality check before you use the result
- Replay the opening, closing and any unclear sections.
- Check every name, number, date and quoted statement you plan to publish.
- Confirm speaker labels match the recording.
- Watch subtitles in the destination player, not only in the transcript editor.
- Keep the original, reviewed transcript and final export clearly named.
- Remove sensitive details before sharing a transcript more widely.
For a longer file, use the first short test to estimate both credit use and review time. A transcript that arrives quickly can still need substantial correction. Budget for that work.
When I would use something else
I would use an existing caption tool if it already fits the editing workflow and produces a draft that is straightforward to correct. I would get a person to review any transcript where an error could have serious consequences. If you need a live integration, use the relevant Realtime documentation rather than this upload guide.
My recommendation: test one short recording
Try a clip you can check easily. Review the words, export your preferred format and judge the full workflow before paying for more capacity.
Try ElevenLabs free →Affiliate link · Mike Says earns commission from qualifying purchases, at no extra cost to you.
Related reading
For voice generation rather than transcription, read the ElevenLabs review. Browse more practical tutorials in Guides.
Sources and update note
Checked 8 October 2026: Speech to Text upload guide, transcript editor and export guide, Speech to Text product and pricing page, and plan pricing. Interface labels, limits and allowances can change. Confirm them in your account before a large upload.