Official documentation
Embedded speech performance evaluations (Speech service)
Microsoft Learn, n.d.
Read the originalRead on October 17, 2026
- What it is
- Microsoft's guide for developers who want to know whether a device can run an embedded speech to text model. It defines the real-time factor (RTF) and gives a worked example and a threshold for real-time use.
- How much weight it carries
- Official vendor documentation, so it is authoritative about Microsoft's own metric and thresholds. It describes one vendor's embedded engine and its own measurement advice, not a neutral benchmark of all recognizers.
Facts taken from this source
- In Microsoft's worked example, a device that processes a one-minute audio file in 30 seconds has a real-time factor of 0.5.
“if a device processes a 1-minute audio file in 30 seconds, the RTF is 0.5” (Metrics & terminology)
- Microsoft defines the real-time factor (RTF) of a device as the ratio of its processing time to the audio length.
“the ratio of the processing time to the audio length” (Metrics & terminology)
- Microsoft says a device should have a real-time factor of 1 or lower to support real-time speech experiences.
“the device should have an RTF of 1 or lower” (Metrics & terminology)