Speech to text benchmark
6 dictation apps, the same 36 talk clips, one Mac. Updated 19 November 2026.
How we test- 95.0of 100 words right, Hushpen with TurboHigher is better
- 0.48 suntil your text appears, Hushpen with Parakeet v3Lower is better
- 17.0 sto write out a 30 minute meetingLower is better
Leaderboard
| Nemotron1 | —3 | ||
|---|---|---|---|
| Parakeet v3 | |||
| Parakeet Unified | —3 | ||
| Parakeet v3 | |||
| Base | |||
| Small | |||
| Turbo | |||
| Turbo | |||
| Turbo | |||
| Turbo2 |
- Nemotron types as you speak. Its last words appear 0.30 s after you stop.
- Buzz loads the app and model for every file, about 13 s.
- Handy and Hushpen Nemotron Unified were not run on 30 minute recordings.
- Small and Base wrote Welsh or Korean on 6 and 8 of 36 clips (language left on Detect).
Our picks
- Near the top, and fasterTurbo
95.0 of 100 words right
0.1 below the top score. Faster than every app that scores as high.
- Fast and accurateParakeet v3
0.48 s until your text appears
93.6 of 100 words right. The default model.
- Accurate and quickParakeet v3 + S1-mini
94.5 of 100 words right
With S1-mini cleanup. Text in 0.94 s.
How long you wait after you stop talking
Hushpen only. The full app typing into TextEdit, middle of 3 runs.
| Nemotron | 0.30 s0.29–0.33 s · 3 runs | 0.79 s0.74–0.89 s · 4 runs | 1.7 s |
|---|---|---|---|
| Parakeet v3 | 0.48 s0.46–0.51 s · 3 runs | 0.94 s0.94–1.01 s · 3 runs | 0.6 s |
| Turbo | 1.69 s1.68–1.69 s · 3 runs | 2.17 s2.15–2.22 s · 3 runs | 0.8 s |
S1-mini loads in 1.1 s on first use.
By accent
Words right out of 100, 19 English accents. Higher is better.
| IndianArunachalam Muruganantham, Sugata Mitra | 89 | 88 | 85 | 90 | 89 | 90 | 85 | 90 |
|---|---|---|---|---|---|---|---|---|
| Spanish or Latin AmericanLuis von Ahn, Pia Mancini | 90 | 87 | 89 | 90 | 90 | 90 | 89 | 90 |
| ChineseEric X. Li, Yang Lan | 92 | 93 | 92 | 91 | 91 | 94 | 93 | 92 |
| AmericanBrené Brown, Simon Sinek | 92 | 94 | 92 | 93 | 94 | 94 | 92 | 94 |
| RussianGarry Kasparov | 93 | 93 | 97 | 93 | 90 | 90 | 93 | 92 |
| JapaneseShigeru Ban, Takaharu Tezuka | 94 | 87 | 86 | 95 | 95 | 88 | 85 | 39 |
| IrishBono, Mary Robinson | 94 | 94 | 93 | 94 | 93 | 93 | 93 | 94 |
| South AfricanNeil Turok, Toby Shapshak | 95 | 90 | 92 | 95 | 95 | 93 | 91 | 94 |
| GermanRalph Langner, Stefan Sagmeister | 95 | 95 | 94 | 95 | 97 | 94 | 95 | 94 |
| BrazilianEduardo Paes, Miguel Nicolelis | 96 | 95 | 93 | 94 | 96 | 94 | 92 | 94 |
| ScottishEvelyn Glennie, Gordon Brown | 96 | 96 | 93 | 96 | 96 | 98 | 95 | 98 |
| KoreanHyeonseo Lee, Joseph Kim | 96 | 95 | 97 | 96 | 96 | 96 | 95 | 95 |
| FrenchEsther Duflo, Matthieu Ricard | 97 | 97 | 96 | 97 | 97 | 97 | 96 | 97 |
| NigerianChimamanda Ngozi Adichie, Ngozi Okonjo-Iweala | 97 | 95 | 95 | 96 | 96 | 95 | 95 | 96 |
| BritishJamie Oliver, Ken Robinson | 97 | 99 | 99 | 99 | 96 | 99 | 99 | 98 |
| AustralianPeter Singer, Rachel Botsman | 98 | 98 | 97 | 100 | 98 | 98 | 98 | 98 |
| PakistaniKhalida Brohi, Ziauddin Yousafzai | 98 | 97 | 98 | 98 | 98 | 97 | 98 | 97 |
| ArabicRana el Kaliouby, Wael Ghonim | 98 | 98 | 96 | 98 | 98 | 98 | 96 | 97 |
| FilipinoMaria Ressa | 99 | 99 | 99 | 97 | 99 | 99 | 97 | 99 |
Best in the rowWorst in the rowHushpen
Two clips per accent. Small and Base are left out. VoiceInk wrote one Japanese-accent clip in Japanese.
What each app says, and what we measured
| App | What the app says | What we measured |
|---|---|---|
"Faster than any server can answer."Speed "Our own purpose-built speech models deliver fast, accurate dictation."Accuracy and speed | 93.5 of 100 words right. 1 minute in 0.37 s. | |
"Local AI models that transcribe your voice to text with 99% accuracy, almost instantly"Accuracy and speed | 92.1 of 100 words right. 1 minute in 4.34 s. | |
"Get accurate text transcriptions in seconds (up to ~30x realtime)"Accuracy and throughput "MacWhisper now supports Parakeet models (up to 300x realtime speeds)"Throughput | 95.0 of 100 words right. 1 minute in 4.95 s. | |
"Fast and accurate, great overall multilingual pick"Speed and accuracy "Best all-around choice for English dictation"English recommendation "3.99"Word error rate "25.3×"Throughput | 94.7 of 100 words right. 1 minute in 0.50 s. |
Not tested yet
"4x faster than typing"Writing speed
"3.4% word error rate on real-world dictation"Word error rate
"Desktop users now fix 13% fewer dictations."Edit frequency
"Slow dictations, the ones that take more than a second to appear, are down 54%."Latency
"transcription latency was 190ms for Willow Voice"Latency
"95%"Zero-edit rate
Claims as each app states them, seen on 19 November 2026. Their tests differ from ours.
Apple M4 Pro Mac, 19 November 2026. Other versions may differ.