Fact · Speech recognition
Turning off condition_on_previous_text can make the text inconsistent across windows, but makes the model less likely to get stuck in a failure loop.
Checked against its source on October 17, 2026
In the source's own words
“disabling may make the text inconsistent across windows, but the model becomes less prone to getting stuck in a failure loop”
whisper/transcribe.py: decoding defaults and command-line options (openai/whisper), OpenAI, GitHub (openai/whisper), 2026, transcribe() parameters: condition_on_previous_text.
Open the original- What it measures
- By default the previous output is fed to the model as a prompt for the next 30-second window. The command-line flag is --condition_on_previous_text, and it is True by default.
- What it does not show
- The choice is a trade-off the source describes. Change it only if you see repeated or looping text in your transcripts.