Start building →OverviewMultilingualCode-switchingEmotion

Speech-to-Text · Multilingual

Chinese (Mandarin)

Language

Request a new language

Tell us which language you want added to this benchmark.

Benchmarks

X
Y
500ms1.00s1.50s6.0%8.0%10.0%Finalize latency lower is betterCER lower is betterSoniox stt-rt-v5Alibaba Qwen3-ASRElevenLabs Scribe v2 RealtimexAI Grok STTOpenAI GPT TranscribeOpenAI GPT-4o TranscribeDeepgram Nova-3Gladia Solaria-1speko.ai/benchmarks

CER

Character error rate — word scoring mismeasures this language.

Sonioxstt-rt-v5Try →5.5%
AlibabaQwen3-ASR6.0%
ElevenLabsScribe v2 Realtime6.2%
xAIGrok STT6.7%
OpenAIGPT TranscribeTry →8.3%
OpenAIGPT-4o TranscribeTry →9.1%
DeepgramNova-3Try →9.3%
GladiaSolaria-1Try →10.3%
0.0%5.0%10.0%15.0%

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 →178ms · 267ms
OpenAIGPT TranscribeTry →761ms · 921ms
GladiaSolaria-1Try →763ms · 2.27s
OpenAIGPT-4o TranscribeTry →827ms · 1.00s
AlibabaQwen3-ASR827ms · 894ms
Sonioxstt-rt-v5Try →860ms · 2.25s
xAIGrok STT1.07s · 1.25s
ElevenLabsScribe v2 Realtime1.61s · 1.73s
0ms300ms600ms900ms1.20s

p50 p90

Time to first token

lower is better

How long before the first word of the transcript arrives.

DeepgramNova-3Try →1.25s · 4.15s
Sonioxstt-rt-v5Try →2.27s · 4.34s
GladiaSolaria-1Try →2.39s · 3.82s
ElevenLabsScribe v2 Realtime2.43s · 4.23s
xAIGrok STT2.53s · 6.93s
OpenAIGPT-4o TranscribeTry →5.63s · 8.82s
OpenAIGPT TranscribeTry →6.71s · 8.98s
AlibabaQwen3-ASR7.15s · 9.40s
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

$2.00
$2.10
$4.50
$4.80
$6.00
$6.50
$12.50
Sonioxstt-rt-v5AlibabaQwen3-ASROpenAIGPT TranscribeDeepgramNova-3OpenAIGPT-4o TranscribeElevenLabsScribe v2 RealtimeGladiaSolaria-1