Vendor page
Measuring STT latency (Deepgram docs)
Deepgram, Deepgram, n.d.
Read the originalRead on October 17, 2026
- What it is
- Deepgram's guide to measuring streaming latency, which defines transcript latency and end-of-turn latency, and states a latency figure for its Nova-3 model.
- How much weight it carries
- Vendor page, so it is reliable for how Deepgram defines and measures latency and for Deepgram's own claims, not for how its service compares with others. The Nova-3 figure is a marketing claim, not an independent measurement.
Facts taken from this source
- Deepgram defines end-of-turn latency as the time from when a user stops speaking to when an end-of-turn event is received.
“The time from when a user stops speaking to when an EOT event is received.” (Latency metrics)
- Deepgram says its Nova-3 model delivers sub-300 ms streaming latency.
“delivering sub-300 ms streaming latency with industry-leading accuracy” (Model considerations)
- Deepgram defines transcript latency as how far behind the transcription is from the audio being sent.
“How far behind the transcription is from the audio being sent.” (Latency metrics)