Preprint
On Word Error Rate Definitions and their Efficient Computation for Multi-Speaker Speech Recognition Systems
Thilo von Neumann et al., arXiv (eess.AS), 2022
Read the originalRead on October 17, 2026
- What it is
- A 2022 paper that gives a single framework for computing WER when a system outputs several speakers' word sequences, and shows that the common multi-speaker WER definitions are special cases of it.
- How much weight it carries
- A preprint from researchers working on multi-speaker recognition. It is a technical paper for specialists, and it describes definitions rather than measuring accuracy on dictation.
Facts taken from this source
- In 2022, von Neumann and colleagues proposed a general framework for computing WER when a recognizer processes recordings with several speakers and outputs several word sequences.
“We propose a general framework to compute the word error rate (WER) of ASR systems that process recordings containing multiple speakers” (Abstract)