Fact · Speech recognition
100frames per second
With a 10-millisecond stride, Whisper's spectrogram has 100 frames for every second of audio.
Checked against its source on October 17, 2026
- What it measures
- Each frame is one slice of the spectrogram: the energy at each frequency over a 25 ms window.
- How it is calculated
- 1 second / 0.010 second stride = 100 frames
- What it does not show
- Counts frames at the Whisper stride. Other systems choose different strides.