Other
jiwer: Evaluate your speech-to-text system (README)
Jitsi open source project, GitHub, jitsi/jiwer, n.d.
Read the originalRead on October 17, 2026
- What it is
- The README of jiwer, an open-source Python package that computes word error rate and related measures for speech-to-text output, with an example of its use.
- How much weight it carries
- An open-source project README that documents the package's behavior. It describes the measures, not an independent test, and the version history of the package changes its internals over time.
Facts taken from this source
- The jiwer README says that applying an empty reference and hypothesis pair to the WER formula is undefined behaviour, because it divides by zero.
“There is undefined behaviour when you apply an empty reference and hypothesis pair to the WER formula, as you get a division by zero.” (Section: A note on empty references)
- The jiwer README says that when the hypothesis is non-empty and the reference is empty, every word or character in the hypothesis counts as an insertion.
“When the hypothesis is non-empty, every word or character counts as an insertion” (Section: A note on empty references)
- The jiwer README offers pip as the alternative to uv for installing the package.
“Or, if you prefer old-fashioned pip and you're using Python” (Section: Installation (before Usage))
- The jiwer Python package computes its speech measures, including WER, from the minimum edit distance between the reference and the hypothesis sentences.
“These measures are computed with the use of the minimum-edit distance between one or more reference and hypothesis sentences” (README, introduction)