Speech-to-Text · Multilingual

Spanish

Language

Request a new language

Tell us which language you want added to this benchmark.

Benchmarks

X
Y

WER

Share of words the transcript got wrong.

AssemblyAIAssemblyAIUniversal-3.6 ProTry →2.53%
AssemblyAIAssemblyAIUniversal-3.5 ProTry →2.8%
ElevenLabsScribe v2 RealtimeTry →3.02%
OpenAIGPT-4o TranscribeTry →3.11%
NariNariQwen3-ASRTry →3.12%
GoogleChirp 33.56%
Sonioxstt-rt-v5Try →3.62%
OpenAIGPT-4o-mini TranscribeTry →4.14%
GradiumASRTry →4.3%
ModulateModulateVelma 2 multilingualTry →4.46%
MetaMetaMuse Voice Transcribe 1.0Try →4.94%
DeepgramNova-3Try →5.24%
Smallest AISmallest AIPulseTry →5.35%
AlibabaQwen3-ASR Realtime6.6%
CartesiaInk-WhisperTry →7.68%
DeepgramNova-2Try →9.06%
0.0%5.0%10.0%15.0%20.0%

Finalize latency

lower is better

Time from the caller stopping to the finalized transcript, not the first partial — the agent cannot reply until it lands.

DeepgramNova-3Try →272ms · 533ms
DeepgramNova-2Try →288ms · 416ms
Sonioxstt-rt-v5Try →478ms · 1.13s
AssemblyAIAssemblyAIUniversal-3.5 ProTry →650ms · 1.67s
Smallest AISmallest AIPulseTry →876ms · 983ms
OpenAIGPT-4o-mini TranscribeTry →898ms · 996ms
OpenAIGPT-4o TranscribeTry →910ms · 1.15s
AlibabaQwen3-ASR Realtime936ms · 1.09s
GradiumASRTry →1.12s · 1.32s
GoogleChirp 31.12s · 1.50s
ElevenLabsScribe v2 RealtimeTry →2.00s · 2.25s
0ms300ms600ms900ms1.20s

p50 p90

Time to first token

lower is better

How long before the first word of the transcript arrives.

AssemblyAIAssemblyAIUniversal-3.5 ProTry →1.77s · 2.28s
DeepgramNova-2Try →2.12s · 3.17s
DeepgramNova-3Try →2.12s · 3.03s
ElevenLabsScribe v2 RealtimeTry →2.21s · 3.16s
Sonioxstt-rt-v5Try →2.22s · 2.91s
Smallest AISmallest AIPulseTry →2.39s · 3.60s
GradiumASRTry →2.73s · 3.33s
GoogleChirp 36.23s · 7.41s
OpenAIGPT-4o TranscribeTry →6.96s · 10.7s
OpenAIGPT-4o-mini TranscribeTry →7.33s · 10.7s
AlibabaQwen3-ASR Realtime7.42s · 8.46s
0ms2.00s4.00s6.00s8.00s

p50 p90

Cost

lower is better

$ per 1,000 minutes of audio · vendor list price, the same in every language