Fact · Languages and accents
The FLEURS paper says that Japanese and Korean transcripts may or may not contain spaces, depending on the transcriber, so the benchmark needed its own handling of spacing.
Checked against its source on October 17, 2026
In the source's own words
“Depending on the transcribers, Japanese and Ko- rean may or may not contain space irregularly.”
FLEURS: Few-shot Learning Evaluation of Universal Representations of Speech (full text), Alexis Conneau et al., arXiv (IEEE Spoken Language Technology Workshop 2022), 2022, Section 2, transcription and tokenization.
Open the original- What it measures
- The authors explain that Chinese text has no spaces between tokens, while Japanese and Korean spacing depends on who transcribed the audio. The paper says it took steps to make comparisons between researchers reproducible.
- What it does not show
- The quote is from the paper's text as published in its PDF, where the word Korean is split across a line break. The sentence describes FLEURS transcripts, not every Japanese or Korean text.