Hallucination (in transcription)
In speech recognition, hallucination is when the model outputs fluent text that was never spoken — typically repeated phrases or invented sentences appearing during silence, music or noise.
Hallucination is a known failure mode of large speech models like Whisper, concentrated in long non-speech stretches. Pipelines defend against it with voice activity detection, silence trimming and repetition filters.
If a transcript contains a suspiciously fluent sentence during what you remember as silence, that is what you are looking at — spot-check against the audio.
Related terms
Frequently asked
What is Hallucination?
In speech recognition, hallucination is when the model outputs fluent text that was never spoken — typically repeated phrases or invented sentences appearing during silence, music or noise.
Why does Hallucination matter?
Hallucination is a known failure mode of large speech models like Whisper, concentrated in long non-speech stretches. Pipelines defend against it with voice activity detection, silence trimming and repetition filters.
What terms are related to Hallucination?
Closely related concepts: Voice activity detection (VAD), Automatic speech recognition (ASR), Confidence score — each has its own entry in this glossary.
Transcribe audio or video in 99+ languages — speaker labels, word timestamps, captions.
Transcribe a file →