Start building →OverviewMultilingualCode-switchingEmotion

Speech-to-Text · Multilingual

Filipino

Language

Request a new language

Tell us which language you want added to this benchmark.

Benchmarks

X
Y
50ms100ms200ms500ms1.00s2.00s10.0%20.0%30.0%Finalize latency lower is better · log scaleWER lower is betterOpenAI GPT-4o TranscribeOpenAI GPT TranscribeOpenAI GPT-4o-mini TranscribeElevenLabs Scribe v2 RealtimeOpenAI GPT Live TranscribeModulate Velma 2 multilingualCartesia Ink-WhisperSoniox stt-rt-v5Deepgram Nova-3Nari Qwen3-ASR FastGoogle Chirp 3Alibaba Qwen3-ASR Realtimespeko.ai/benchmarks

WER

Share of words the transcript got wrong.

ModulateModulateVelma 2 multilingualTry →6.2%
Sonioxstt-rt-v5Try →8.9%
MetaMetaMuse Voice Transcribe 1.0Try →9.7%
ElevenLabsScribe v2 RealtimeTry →11.7%
OpenAIGPT-4o TranscribeTry →12.3%
OpenAIGPT Live TranscribeTry →12.5%
OpenAIGPT TranscribeTry →13.4%
CartesiaInk-WhisperTry →17.2%
OpenAIGPT-4o-mini TranscribeTry →17.5%
DeepgramNova-3Try →24.5%
NariNariQwen3-ASR FastTry →26.4%
GoogleChirp 326.6%
NariNariQwen3-ASRTry →26.7%
AlibabaQwen3-ASR RealtimeTry →29.7%
0.0%10.0%20.0%30.0%40.0%

Finalize latency

Region

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

NariNariQwen3-ASR FastTry →34ms · 57ms
CartesiaInk-WhisperTry →302ms · 498ms
Sonioxstt-rt-v5Try →337ms · 1.18s
ModulateModulateVelma 2 multilingualTry →402ms · 1.28s
GoogleChirp 3646ms · 901ms
DeepgramNova-3Try →707ms · 1.17s
OpenAIGPT-4o TranscribeTry →806ms · 1.03s
OpenAIGPT-4o-mini TranscribeTry →845ms · 1.01s
OpenAIGPT TranscribeTry →853ms · 1.10s
AlibabaQwen3-ASR RealtimeTry →1.06s · 1.21s
ElevenLabsScribe v2 RealtimeTry →1.94s · 2.13s
OpenAIGPT Live TranscribeTry →2.37s · 2.91s
0ms300ms600ms900ms1.20s

p50 p90

Time to first token

Region

How long before the first word of the transcript arrives.

NariNariQwen3-ASR FastTry →1.74s · 1.86s
DeepgramNova-3Try →2.15s · 3.14s
ElevenLabsScribe v2 RealtimeTry →2.29s · 3.19s
Sonioxstt-rt-v5Try →2.52s · 3.01s
ModulateModulateVelma 2 multilingualTry →2.55s · 3.02s
OpenAIGPT Live TranscribeTry →2.61s · 3.18s
GoogleChirp 33.98s · 6.28s
CartesiaInk-WhisperTry →5.10s · 6.94s
OpenAIGPT-4o TranscribeTry →5.20s · 6.89s
OpenAIGPT TranscribeTry →5.36s · 7.08s
OpenAIGPT-4o-mini TranscribeTry →5.41s · 7.19s
AlibabaQwen3-ASR RealtimeTry →6.83s · 7.83s
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

$1.00
$2.00
$2.00
$2.10
$3.00
$3.70
$4.50
$4.80
$5.40
$6.00
$6.50
$16.00
$17.00
NariNariQwen3-ASRNariNariQwen3-ASR FastSonioxstt-rt-v5AlibabaQwen3-ASROpenAIGPT-4o-mini TranscribeElevenLabsScribe v2OpenAIGPT TranscribeDeepgramNova-3AlibabaQwen3-ASR RealtimeOpenAIGPT-4o TranscribeElevenLabsScribe v2 RealtimeGoogleChirp 3OpenAIGPT Live Transcribe