Speaker identification
Scribewren separates the voices in a recording and labels each turn as Speaker 1, Speaker 2 and so on. Rename a speaker once and the name applies everywhere.
The name applies to every segment of this speaker.
- Priya Nair 02:10
Before we begin, a quick word on how the session will run.
- Daniel 02:18
Sounds good. I will keep my answers short.
- Priya Nair 02:24
Thank you. First question: what changed in the second year?
- Daniel 02:31
Mostly the team. We hired people who had run field studies.
What it does
Speaker labels are added to your transcriptions on every plan. Each time the speaker changes, the transcript starts a new segment with that speaker label.
Scribewren tells voices apart inside one file. It does not match a voice across files. It does not know who anyone is. It does not keep voiceprints. You supply the real names.
Labels are less certain when voices are alike, when the audio is poor or when people talk over each other. Overlapping speech goes to the speaker who dominates it, so check those turns by ear.
- Segments break where the speaker changes.
- Rename a speaker once and every segment and every export uses the new name.
- Move a segment to another speaker, or to a new one, from the label menu.
- Export with speaker names, or leave them out.
How it works
- Upload a recording. Speaker labels are added without any setting to change.
- Open the transcript and read the turns, marked Speaker 1, Speaker 2 and so on.
- Click a label and choose Rename speaker to give the voice a real name.
- If a turn is wrong, use the same menu to move it to another speaker or add a new one.
- Export with the speaker names included, or without them.
Details
- Labels
- Speaker 1, Speaker 2 and so on, until you rename them.
- Renaming
- One rename covers the whole transcript and its exports.
- Scope
- Inside one file. No voice matching across files and no voiceprints.
- Overlapping speech
- Assigned to the dominant speaker.
- Very short files
- Files under 10 seconds skip speaker detection.
- If it fails
- You still get the transcript, with a notice that speaker detection was not available for the file.
Plan availability
| Free | Unlimited | |
|---|---|---|
| Speaker labels | Included | Included |
Speaker labels are the same on every plan. Compare the plans.
More features
Questions
Can I tell Scribewren how many speakers there are?
No. Speakers are detected automatically, and you can fix any turn by hand in the editor.
Does Scribewren identify who a person is?
No. It tells voices apart inside one file and calls them Speaker 1, Speaker 2 and so on. You add the names. It does not recognise people across files and does not store voiceprints.
Can I turn speaker labels off?
Speaker detection runs on its own and has no switch. When you export, you can leave the speaker names out of the file.
Does it work for group recordings?
Interviews, panels and calls work well when people take turns. Large groups, similar voices and crosstalk make the labels less certain, so expect to correct some turns.
Is speaker identification included on Free?
Yes. Speaker labels come with every plan.
Try it on your own recording
Free includes a few transcriptions a day. No card needed. Compare plans.
Start free