By language
Accuracy, language by language.
Most notetakers publish one accuracy figure, or none at all. These are our measured error rates per language, against the industry-standard engine behind most competing products.
| Language | Canto | Industry-standard engine | Difference | |
|---|---|---|---|---|
| SerbianSrpski | 4.4% | 92.2% | 21× fewer | See the numbers → |
| BosnianBosanski | ≤5% | 16.6% | 3.3× fewer | See the numbers → |
| CroatianHrvatski | ≤5% | 14.5% | 2.9× fewer | See the numbers → |
| PolishPolski | 2.3% | 5.9% | 2.6× fewer | See the numbers → |
| GermanDeutsch | 2.7% | 4.5% | 1.7× fewer | See the numbers → |
Word error rate counts every insertion, deletion and substitution against a human reference transcript, then divides by the number of words spoken. Lower is better, and 5% means roughly one wrong word in twenty. We measure on FLEURS, a public benchmark of read speech across many languages, so anyone can repeat the test rather than take our word for it.