Fact · Speech recognition
860milliseconds (90th percentile endpointer latency)
Sainath et al. report that with an acoustic endpointer alone, the 90th-percentile endpointer latency was 860 ms.
Checked against its source on October 17, 2026
In the source's own words
“With the acoustic endpointer alone, we degrade the WER from 6.2% (no EP) to 7.4% to achieve a 450ms EP50 latency and 860ms EP90 latency.”
A Streaming On-Device End-to-End Model Surpassing Server-Side Conventional Model Quality and Latency, Tara N. Sainath et al., ICASSP 2020 (arXiv:2003.12710), 2020, Section 6.2.1, Table 4.
Open the original- What it measures
- Same experiment and test set as the 450 ms median figure.
- What it does not show
- The paper reports the 90th percentile as EP90, and one experiment on one system.