CekuraBench
Speech to textText to speech
Book a demo
All benchmarks›Vendors›Fish Audio

Fish Audio

1 model across speech to text.

Last updated Nov 3, 2026 · Methodology by Dileep Chagam

Building on Fish Audio?

Test your agent the same way.

Book a demoStart free

Speech to text

Full results →
Transcribe Stream 12.21% WER3rd of 24

Other vendors

  • OpenAI
  • Google
  • xAI
  • AssemblyAI
  • Deepgram
  • Amazon
  • Boson AI
  • Deepslate
  • Microsoft
  • Phonic
  • Reson8
  • Cartesia
  • ElevenLabs
  • Gradium
  • Inworld
  • Murf
  • NVIDIA
  • PolyAI
  • Rime
  • Smallest
  • Soniox
  • Speechmatics
  • Zoom
Cekura

Public benchmark results for voice agent platforms, realtime speech models, and speech recognition systems.

Speech to speech
Agent performanceTurn benchReal human calls
Speech and voice
Speech to textText to speechVoice platformsVoice quality
Explore
All benchmarksCompare S2S modelsCompare STT modelsCompare platformsVendorsExperiments
About
About the benchmarkGovernance and changelogSource repositoryDataset licenseCekura websiteDocumentationBlogContact
© 2026 CekuraTermsPrivacy
Fish Audio · 1 model benchmarked