Fact · Speech recognition
480000samples per 30-second chunk
In the open-source Whisper code, a chunk of 30 seconds at a sample rate of 16,000 per second is 480,000 samples.
Checked against its source on October 17, 2026
In the source's own words
“N_SAMPLES = CHUNK_LENGTH * SAMPLE_RATE # 480000 samples in a 30-second chunk”
whisper/audio.py: the audio constants and log-Mel code in the openai/whisper repository, OpenAI, GitHub, openai/whisper repository (main branch), 2022, Module constants, audio.py.
Open the original- What it measures
- The constants sit at the top of the file that converts audio into the spectrogram the model reads.
- What it does not show
- The number is the code's own comment for the open-source Whisper release. It does not say how long a real recording takes to transcribe.