Live and real-time transcription

Daisy can show the transcript as you record. How quickly it updates depends on your provider and a couple of settings. This article explains the options and what to expect.

Behavior settings

Open Settings and go to the Behavior section:

• Show transcript live in the recording window — turns the live transcript on or off during a recording.
• Stream audio to Deepgram for sub-second transcript updates — enables real-time transcription. This requires Deepgram and is available only when the live transcript is on.
• Chunked segment length (seconds) — shown when real-time is off. Sets how often audio is sent for transcription. The range is 5 to 60 seconds; the default is 15.

Real-time vs. chunked

There are two ways the live transcript updates:

• Word-by-word (real-time) — requires Deepgram with streaming enabled. The transcript updates as people speak, with sub-second latency.
• Semi-live (chunked) — used with any other provider, or when streaming is off. Daisy sends audio in segments and shows the transcript a chunk at a time, every 15 seconds by default.

If real-time is on but Deepgram is not configured, the transcript falls back to chunked segments at your configured interval.

What you see while recording

The live transcript labels speech by source: your microphone shows as Me and other participants show as Them. Interim text — words still being finalized — appears slightly faded and in italics until it settles, so you can follow along before each segment is locked in.

Good to know

Lower chunk lengths feel more responsive but send audio more often. The default of 15 seconds is a balance between responsiveness and overhead for chunked transcription.