Question · Languages and accents
How accurate is speech recognition in Estonian?
On our Mac benchmark, Estonian read speech is transcribed with a word error rate of 14.17 percent by Parakeet and 19.91 percent by Whisper large-v3-turbo. That ranks Estonian twentieth of the 25 languages tested.
Updated October 19, 20263 min read11 cited facts, each traced to its source
Estonian sits in the lower middle of the languages we tested: about 1 word in 7 is wrong with the better engine. In our October 2026 benchmark on a Mac with an M2 Pro, Parakeet TDT 0.6B v3 had a word error rate of 14.17 percent 1 and Whisper large-v3-turbo had 19.91 percent 2, on 100 Estonian recordings from the FLEURS test set. We have no Apple SpeechTranscriber result for Estonian.
Parakeet is clearly better than Whisper in Estonian
Parakeet beats Whisper by several points of word error rate, a gap that is large relative to the sample noise. It is also faster: about 22 times faster than real time 3, against about 15 times for Whisper 4. In Estonian it is better and quicker. The speed dataset covers every language.
Our measurement of the engines in Estonian
Lower is better in this table.
| Engine | Word error rate (WER %) | Source |
|---|---|---|
| Parakeet TDT v3 | 14.17 | Calculated 1 |
| Whisper large-v3-turbo | 19.91 | Calculated 2 |
The transcripts behind each row are on the on-device benchmark page.
Published scores differ by test, but the order holds
NVIDIA reports 17.73 percent for Parakeet on FLEURS Estonian 5, higher than our 14.17 percent. OpenAI's paper gives 21.9 percent for the older Whisper large-v2 6. On CoVoST 2 Estonian, Parakeet had 22.04 percent 7. In every set the figures sit well above those for Spanish or Italian, and the exact number depends on the recordings.
| System | Kind | Test set | Speech | Word error rate (WER %) | Source |
|---|---|---|---|---|---|
| Parakeet TDT 0.6B v3 | Open model | FLEURS | Clean read speech | 14.17 | Calculated 1 |
| NVIDIA Parakeet TDT 0.6B v3 | Open model | FLEURS | Clean read speech | 17.73 | NVIDIA, 2025 5 |
| Whisper large-v3-turbo | Open model | FLEURS | Clean read speech | 19.91 | Calculated 2 |
| Whisper large-v2 | Open model | FLEURS | Clean read speech | 21.9 | Radford et al., 2022 6 |
| NVIDIA Parakeet TDT 0.6B v3 | Open model | CoVoST 2 | 22.04 | NVIDIA, 2025 7 |
All published rows are on the published accuracy page.
Estonian ranks twentieth of 25 languages
Taking the best engine in each language, Estonian is twentieth of the 25 languages in the benchmark. Danish is just behind it, and Latvian (17.27 percent 8) and Lithuanian are further back. Finnish, which is closely related to Estonian, is much better at 7.73 percent 9. Estonian has far fewer speakers and much less training audio than Finnish, though the benchmark does not test the cause.
Formatting adds little for Parakeet and more for Whisper
Counting punctuation and capital letters as a reader sees them, Parakeet's text differed from the written sentences by 15.29 errors per 100 tokens 10, and Whisper's by 20.52 11. The formatting penalty is small in Estonian, so the word errors dominate what you will see.
What to expect when you dictate in Estonian
Expect to correct about 1 word in 7 with Parakeet and about 1 in 5 with Whisper, even in a quiet room. Dictation is a useful first draft that you proofread, not finished text. Choose Parakeet if your tool offers it. Estonian words change shape with many case endings, and a wrong ending counts as a whole wrong word, so names and long compound words are the likely trouble spots. We found no published fact that confirms Estonian in a built-in Mac or Windows dictation list, so check your system's language list, and read our overview of dictation languages before you choose a tool.
What these numbers do not show
The recordings are people reading prepared sentences in a quiet room, scored on one Mac with 100 recordings. Spontaneous speech, dialects, noise and specialist vocabulary are not in the test, and we did not measure Apple's model or any cloud service ourselves. Real dictation will usually score worse than read speech. For the measure itself, see the guide on how accurate speech recognition is.
Our guide to European languages in speech to text helps you choose an engine, and the languages overview covers other languages. For comparison, read the same question for Finnish, Latvian and Danish.
Evidence
Every number on this page, with where it comes from. Open a fact to see the exact words in its source.
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
The data behind these numbers
- Speech Recognition Accuracy: Word Error Rates7 of this page's figures, 673 data points in all
- On-Device Speech Recognition Benchmark: 25 Languages4 of this page's figures, 56 data points in all
- Speech Recognition Speed: Real-Time Factor Data2 of this page's figures, 66 data points in all
- Dictation Punctuation and Capitalization Accuracy2 of this page's figures, 56 data points in all
Go deeper
Related questions
- Does dictation work for both Brazilian and European Portuguese?
- Does speech recognition understand Swiss German?
- How accurate is speech recognition in Bulgarian?
- How accurate is speech recognition in Croatian?
- How accurate is speech recognition in Czech?
- How accurate is speech recognition in Danish?
- How accurate is speech recognition in Dutch?
- How accurate is speech recognition in English?