Cantonese Voice Benchmark粵語語音基準
EN繁中

Cantonese Voice Benchmark 2026.1 · Window 2

Leaderboard

Every published number carries a bootstrap confidence interval. Entries that overlap are reported as tied rather than ordered — with 30 clips a fraction of a percent is noise.

Window
Window 2
Closes
02 Aug, 14:21 HKT
Entrants
0
Updated
—
Ranking
Core ASR + LLMFull all pillars
Difficulty
AllEasyNormalHard

No results published yet

The first scored window has not closed. Reference baselines are shown below so the scale is meaningful the moment real entries land.

A
Open-source pipeline
71.0
11.1%
3.81
3.44
3.29
2.11sp95 3.88
B
Proprietary voice LLM
78.4
8.6%
4.14
3.62
4.01
0.88sp95 1.52
Reference baselines
AOpen-source pipeline71.0
BProprietary voice LLM78.4
How we measureNo snapshot yet · scoring v1.2
Character error rate with 95% confidence intervalsOpen-source pipeline: 11.1 percent. Proprietary voice LLM: 8.6 percent9.0%10.0%11.0%Open-source pipeline11.1Proprietary voice LLM8.6
Reference systems only so far. Bands appear once two entrants have published intervals to compare.
TiedOverlapping 95% confidence intervals. Reported as tied.
Not enteredDid not compete on that pillar. Never a zero.
PendingAwaiting human evaluation.
Insufficient dataUnder 80% of items completed without a platform fault. No score, no quota consumed.