Pipecat · Whisper-tiny

audio · prosody

* Conditional

Eval set is Smart Turn’s home distribution — treat 94% as a ceiling.

EOT accuracy

94.0%

END-recall

89.0%

False-cutoff

1.0%

One false cutoff in 100 incomplete clips.

Inference

32ms

49 p90

Runs in parallel with STT — no transcript wait.

Share this result

Download card (PNG) Post on X

Embed the live badge

Markdown

[![Speko turntaking rank](https://benchmarks.speko.ai/badge/turntaking/smart-turn-v3-2-pipecat-whisper-tiny.svg)](https://benchmarks.speko.ai/turntaking/smart-turn-v3-2-pipecat-whisper-tiny)

HTML

<a href="https://benchmarks.speko.ai/turntaking/smart-turn-v3-2-pipecat-whisper-tiny"><img src="https://benchmarks.speko.ai/badge/turntaking/smart-turn-v3-2-pipecat-whisper-tiny.svg" alt="Speko turntaking rank"></a>

URL

https://benchmarks.speko.ai/turntaking/smart-turn-v3-2-pipecat-whisper-tiny