AssemblyAI STT 3.7 Realtime Agent vs GPT-4o Mini Transcribe

Compare

Accuracy and speed

Left gets fewer words wrong. Lower has the final text ready sooner.

Accurate and fast0ms200ms400ms600ms800ms0%2%4%6%Words wrong on clean audio (Pipecat WER)
  • STT 3.7 Realtime Agent1.68% · 47ms
  • 4o Mini Transcribe3.35% · 630ms

AssemblyAI STT 3.7 Realtime Agent is ahead of GPT-4o Mini Transcribe on both datasets together, the Pipecat Dataset and final-text delay. The Ocular Dataset is too close to call.

Last updated Methodology by Dileep Chagam

Who is better at what

STT 3.7 Realtime Agent4o Mini Transcribe
  • Overall WER2.41%3.73%
  • Pipecat Dataset WER1.68%3.35%
  • Ocular Dataset WER3.81%4.45%
  • Time to first text493msn/a
  • Final-text delay47ms630ms

Lower is better on every measure. Too close to call means the likely ranges overlap.

Clip by clip

On every clip both models were scored on, the lower WER wins, and WERs within 5 points of each other are a draw; in the rating, a wider win counts for more. Expected win rate is what the two ELO ratings predict for a match.

STT 3.7 Realtime Agent4o Mini Transcribe
  • ELO rating10241001–1039 · rank 1–161001993–1011 · rank 6–18
  • Expected win rate53%47%
  • Ocular turns WER · 162 draws of 19725 won10 won
  • Pipecat WER · 888 draws of 99974 won37 won
  • Semantic WER · 185 draws of 1976 won6 won
  • Phone line · 166 draws of 19822 won10 won
  • Call-center noise · 126 draws of 19751 won20 won
  • Babble noise · 112 draws of 19763 won22 won
  • TV noise · 112 draws of 19518 won65 won
  • Reverb · 140 draws of 19846 won12 won
  • Short replies · 56 draws of 7010 won4 won
  • Entities · 46 draws of 492 won1 won

Short replies are decided on the error count and entities on exact capture, with no draw margin. How the rating is worked out

STT 3.7 Realtime Agent vs 4o Mini Transcribe FAQ

  • AssemblyAI STT 3.7 Realtime Agent is more accurate on both datasets. On the 1,000 public Pipecat clips, AssemblyAI STT 3.7 Realtime Agent scores 2.41% and GPT-4o Mini Transcribe 3.73%, so AssemblyAI STT 3.7 Realtime Agent makes fewer errors by 1.32 points. On the 300 licensed Ocular turns, AssemblyAI STT 3.7 Realtime Agent scores 1.68% and GPT-4o Mini Transcribe 3.35%, so AssemblyAI STT 3.7 Realtime Agent makes fewer errors by 1.68 points. Word error rate counts incorrect, missing, and extra words, so lower is better.

  • Final-text delay was not measured for one or both, so speed cannot be compared. Both are measured on the same 206 streamed turns.

  • AssemblyAI STT 3.7 Realtime Agent makes fewer errors on real conversations, and the two are not separated on speed. A voice agent cannot reply until the final transcript arrives, and the Ocular recordings are real conversations.

  • Only AssemblyAI STT 3.7 Realtime Agent. GPT-4o Mini Transcribe returns no text until the speaker stops, which matters if your agent reacts to partial transcripts.

Test your voice agent the same way.

Cekura runs these simulated calls on your agent and scores every one.