AssemblyAI STT 3.7 Realtime Agent vs Smallest Pulse

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
  • Pulse3.66% · 201ms

AssemblyAI STT 3.7 Realtime Agent is ahead of Smallest Pulse on both datasets together, the Pipecat Dataset, time to first text 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 AgentPulse
  • Overall WER2.41%3.59%
  • Pipecat Dataset WER1.68%3.66%
  • Ocular Dataset WER3.81%3.46%
  • Time to first text493ms1.06s
  • Final-text delay47ms201ms

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 AgentPulse
  • ELO rating10241001–1039 · rank 1–161005998–1011 · rank 6–18
  • Expected win rate53%47%
  • Ocular turns WER · 155 draws of 19832 won11 won
  • Pipecat WER · 799 draws of 999168 won32 won
  • Semantic WER · 183 draws of 1989 won6 won
  • Phone line · 160 draws of 19727 won10 won
  • Call-center noise · 119 draws of 19651 won26 won
  • Babble noise · 123 draws of 19744 won30 won
  • TV noise · 144 draws of 19832 won22 won
  • Reverb · 158 draws of 19830 won10 won
  • Short replies · 51 draws of 7017 won2 won
  • Entities · 45 draws of 504 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 Pulse 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 Smallest Pulse 3.59%, so AssemblyAI STT 3.7 Realtime Agent makes fewer errors by 1.17 points. On the 300 licensed Ocular turns, AssemblyAI STT 3.7 Realtime Agent scores 1.68% and Smallest Pulse 3.66%, so AssemblyAI STT 3.7 Realtime Agent makes fewer errors by 1.98 points. Word error rate counts incorrect, missing, and extra words, so lower is better.

  • Yes. AssemblyAI STT 3.7 Realtime Agent finishes the transcript 564ms sooner. AssemblyAI STT 3.7 Realtime Agent has the final transcript 493ms after the speaker stops and Smallest Pulse 1.06s, at the median. Both are measured on the same 206 streamed turns.

  • AssemblyAI STT 3.7 Realtime Agent is the stronger pick for voice agents: it makes fewer errors on real conversations and finishes the transcript 564ms sooner. A voice agent cannot reply until the final transcript arrives, and the Ocular recordings are real conversations.

  • Yes, both return partial text before the speaker stops. AssemblyAI STT 3.7 Realtime Agent shows first text after 3.81% and Smallest Pulse after 3.46%, at the median.

Test your voice agent the same way.

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