Today we are rolling out an upgrade to speaker labels in InstantTranscriber.
The goal is simple: when a transcript has multiple people speaking, it should be easier to see who said what, easier to review uncertain moments, and easier to use the transcript in real work.
What changed
Speaker-labeled transcripts now get an additional quality pass when eligible. This pass looks for places where the speaker assignment may be uncertain and improves the final speaker labels when there is enough context to do so.
You will still see your transcript as soon as it is ready. If the additional speaker-label pass is still running, the viewer will show that status and update automatically when it finishes.
Compare the original and improved labels
When the enhanced speaker labels are available, they are shown by default. You can also switch them off in the transcript viewer to compare against the original speaker labels.
This is useful when you are reviewing interviews, meetings, podcasts, sales calls, research recordings, or any other multi-speaker audio where speaker attribution matters.
Speaker names, when the transcript makes them clear
In some recordings, a speaker is named directly in the conversation. When that is clear from the transcript, InstantTranscriber can show that name instead of a generic label like SPEAKER_00.
We intentionally keep this conservative. If a name is not clearly present in the transcript, we keep the normal speaker label instead of guessing.
Better handling for split speaker labels
Sometimes one real speaker can be split into more than one generic label, especially in short clips, noisy audio, or recordings with quick back-and-forth turns. The new quality pass can now review some of those cases and combine labels only when the evidence is strong enough.
As with names, this is designed to be careful. We would rather keep two labels separate than incorrectly merge two different people.
What Free users get
- Standard speaker identification remains available on the Free plan.
- Timestamps and speaker toggles remain available.
- Free users can try the higher-quality speaker-label pass on their first transcription.
- Free plan limits still apply: 3 transcriptions per day, up to 35 minutes and 50 MB per file.
What Premium users get
- Higher-quality speaker-label processing on eligible transcriptions.
- Improved review of uncertain speaker turns.
- Optional speaker names when names are clearly present in the transcript.
- Original/enhanced label comparison in the viewer.
- Premium limits: unlimited transcriptions, up to 10 hours and 1 GB per file.
What stays the same
The transcript text is still ready quickly. The speaker-label quality pass is an additional layer on top of the normal transcription workflow, not a reason to wait before reading or exporting your text.
As always, audio quality matters. Clear recordings, minimal overlap, and distinct voices will produce the best results.
You can try the upgraded speaker-label experience now in your InstantTranscriber dashboard.