Speaker-attributed transcription
Speaker-attributed transcription is a transcript that labels who said each line, not just what was said. The label can come from the audio source, from voice-based diarization or from the meeting app's participant list. Reline uses speaker separation (Me vs Other: your microphone vs the call audio), and a beta option numbers individual voices for you to name.
We're aligned on the rollout date — what's the gating risk on the auth flow?
The refresh-token edge case. Once that's covered we can ship to the pilot cohort.
Send me the failure modes before EOD and I'll bake them into the test matrix.
Will do — pairing with the platform team tomorrow morning to align on the schema...
In depth
Speaker-attributed transcription is a transcript that labels who said each line, rather than one undifferentiated block of text. Attribution turns a transcript into something you can scan and quote: you can see who committed to an action item or raised a concern. Labels can come from voice-based speaker separation, from a split between your microphone and the rest of the call, or from the meeting app's participant list. Reline starts with the split, speaker separation (Me vs Other: your microphone vs the call audio), so in any of its 60+ languages each line shows which side of the call it came from, as long as the call audio is captured. A beta option also numbers individual voices.
In practice
Attribution is what makes a transcript useful after the call. You search for who agreed to send the contract, who pushed back on the date, or what the client actually asked for, and the answer depends on the label beside each line. Labels go wrong in predictable places: people talking over each other, several people around one laptop, a guest on a phone line, or, in tools that split by audio source, a recording where the call audio was never captured, so every line lands on your side. When you compare tools, ask where the labels come from: separate audio streams, voice clustering, or the meeting app's participant list. Each fails differently, so also check how quickly you can fix a wrong label.
How Reline does it
Reline records without a bot, from your computer, and keeps two streams apart: your microphone and the call audio. Each line lands on your side (Me) or the call's side (Other) by the stream it came from, so the split holds even when voices sound alike. Without the call audio, every line lands on your side. To tell people apart within one side, two beta options number the voices: Separate speakers for the call, Shared mic for a room around one microphone. The numbers are an estimate and never names, so you click a voice and name it yourself.
Open a note and press Start. Reline records your microphone and the call audio as two separate streams.
Press Show transcript. Your lines and the call's lines appear on opposite sides, so each line shows which side it came from.
On a call with several people, open Who is on the other side beside the record button and choose Separate speakers.
Click a label such as Speaker 1 and type a name in Name this speaker. It applies to every line from that voice.
Questions
- What is the difference between speaker attribution and speaker diarization?
- Diarization is one method: it groups audio by voice and answers who spoke when, usually with anonymous labels like Speaker 1. Attribution is the result on the page, each line tied to a speaker. That label can come from diarization, from separate audio channels, or from the meeting app's participant list.
- How do you get a transcript with speaker names?
- Names need a source the audio alone does not give. The meeting platform's built-in transcription can use the names people joined with, if your admin allows it. Other tools read the participant list, use voice profiles, or number the voices so you can rename them. Always check names where people talk over each other.
- Does Reline show speaker names in transcripts?
- Not automatically. Reline labels each line Me or Other, from your microphone versus the call audio. A beta option numbers the voices on either side (Speaker 1, Speaker 2), and you add the names yourself: name a voice once and the name applies to everything that voice says in the note.
Stop taking notes.Start shipping outcomes.
Free forever for individuals. Five minutes to install. Your next meeting writes its own notes.