Fact · Speech recognition
12.75percent
In the formatted accuracy of on-device speech recognition: punctuation and capitals counted dataset, the 25 published figures for “Parakeet TDT v3” have a median of 12.75 percent, ranging from 5.44 to 34.83.
Calculated from 25 data points on October 19, 2026
- What it measures
- Each figure is one row of the dataset: a published measurement for a stated group, traced to its source. The median is the middle value of the 25 figures for this group.
- How it is calculated
- Median of 25 values (5.44, 6.4, 7.06, 7.37, 7.68, 8.04, 8.18, 8.42, 9.91, 10.29, 11.07, 12.1, 12.75, 12.93, 14.74, 15.29, 15.52, 15.69, 16.56, 18.54, 18.62, 20.07, 21.81, 25.96, 34.83) = 12.75.
- What it does not show
- FLEURS sentences are read aloud from Wikipedia, with the source's punctuation; optional commas and style choices count as errors for every engine alike, so compare engines within a language rather than reading the number as the editing a person needs.
A statistic of the dataset Dictation Punctuation and Capitalization Accuracy, calculated by code from its published rows.
Calculated from
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Bulgarian sentences by 15.69 errors per 100 tokens, against 13.42 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Czech sentences by 12.75 errors per 100 tokens, against 10.44 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Danish sentences by 18.54 errors per 100 tokens, against 15.78 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written German sentences by 7.06 errors per 100 tokens, against 4.84 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Greek sentences by 34.83 errors per 100 tokens, against 31.66 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written English sentences by 8.18 errors per 100 tokens, against 4.48 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Spanish sentences by 5.44 errors per 100 tokens, against 2.79 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Estonian sentences by 15.29 errors per 100 tokens, against 14.17 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Finnish sentences by 12.10 errors per 100 tokens, against 10.11 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written French sentences by 8.04 errors per 100 tokens, against 5.29 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Croatian sentences by 15.52 errors per 100 tokens, against 12.44 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Hungarian sentences by 16.56 errors per 100 tokens, against 14.09 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Italian sentences by 6.40 errors per 100 tokens, against 2.52 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Lithuanian sentences by 20.07 errors per 100 tokens, against 17.75 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Latvian sentences by 18.62 errors per 100 tokens, against 17.27 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Maltese sentences by 25.96 errors per 100 tokens, against 22.02 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Dutch sentences by 8.42 errors per 100 tokens, against 6.32 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Polish sentences by 11.07 errors per 100 tokens, against 8.21 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Portuguese sentences by 7.68 errors per 100 tokens, against 3.62 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Romanian sentences by 12.93 errors per 100 tokens, against 10.00 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Russian sentences by 7.37 errors per 100 tokens, against 4.74 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Slovak sentences by 10.29 errors per 100 tokens, against 8.45 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Slovenian sentences by 21.81 errors per 100 tokens, against 19.83 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Swedish sentences by 14.74 errors per 100 tokens, against 12.03 for the words alone.
- In our October 2026 benchmark, counting punctuation and capital letters as a reader sees them, Parakeet TDT 0.6B v3's text differed from the written Ukrainian sentences by 9.91 errors per 100 tokens, against 5.65 for the words alone.