Character error rate (CER)
Character error rate (CER) measures speech-recognition accuracy at the character level rather than the word level, and is the standard metric for languages without spaces between words, such as Chinese, Japanese and Thai.
For languages where 'word' is ambiguous, counting character edits gives a more stable measure than WER. For alphabetic languages CER is also useful to distinguish near-misses (one wrong letter) from total misses (a different word entirely).
Related terms
Frequently asked
What is Character error rate?
Character error rate (CER) measures speech-recognition accuracy at the character level rather than the word level, and is the standard metric for languages without spaces between words, such as Chinese, Japanese and Thai.
Why does Character error rate matter?
For languages where 'word' is ambiguous, counting character edits gives a more stable measure than WER. For alphabetic languages CER is also useful to distinguish near-misses (one wrong letter) from total misses (a different word entirely).
What terms are related to Character error rate?
Closely related concepts: Word error rate (WER), Automatic speech recognition (ASR) — each has its own entry in this glossary.
Transcribe audio or video in 99+ languages — speaker labels, word timestamps, captions.
Transcribe a file →