AssemblyAI STT 3.7 Realtime Agent vs Inworld STT-1

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
  • STT-12.72% · 41ms

AssemblyAI STT 3.7 Realtime Agent is ahead on both datasets together and time to first text, and Inworld STT-1 on final-text delay. The Pipecat Dataset and The Ocular Dataset are too close to call.

Last updated Methodology by Dileep Chagam

Who is better at what

STT 3.7 Realtime AgentSTT-1
  • Overall WER2.41%5.44%
  • Pipecat Dataset WER1.68%2.72%
  • Ocular Dataset WER3.81%10.60%
  • Time to first text493ms1.06s
  • Final-text delay47ms41ms

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 AgentSTT-1
  • ELO rating10241001–1039 · rank 1–1610311011–1043 · rank 1–13
  • Expected win rate49%51%
  • Ocular turns WER · 179 draws of 1987 won12 won
  • Pipecat WER · 897 draws of 99963 won39 won
  • Semantic WER · 192 draws of 1981 won5 won
  • Phone line · 180 draws of 1979 won8 won
  • Call-center noise · 139 draws of 19730 won28 won
  • Babble noise · 131 draws of 19734 won32 won
  • TV noise · 137 draws of 19825 won36 won
  • Reverb · 170 draws of 19813 won15 won
  • Short replies · 63 draws of 702 won5 won
  • Entities · 47 draws of 502 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 STT-1 FAQ

  • AssemblyAI STT 3.7 Realtime Agent is more accurate on the public clips. On the 1,000 public Pipecat clips, AssemblyAI STT 3.7 Realtime Agent scores 2.41% and Inworld STT-1 5.44%, so AssemblyAI STT 3.7 Realtime Agent makes fewer errors by 3.02 points. On the 300 licensed Ocular turns, AssemblyAI STT 3.7 Realtime Agent scores 1.68% and Inworld STT-1 2.72%, with likely ranges of 1.47 to 1.90% and 1.67 to 4.68%. The ranges overlap, so the difference is too close to call. Word error rate counts incorrect, missing, and extra words, so lower is better.

  • Yes. AssemblyAI STT 3.7 Realtime Agent finishes the transcript 570ms sooner. AssemblyAI STT 3.7 Realtime Agent has the final transcript 493ms after the speaker stops and Inworld STT-1 1.06s, at the median. Both are measured on the same 206 streamed turns.

  • AssemblyAI STT 3.7 Realtime Agent finishes the transcript 570ms sooner, and the two are too close to call on real conversations. 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 Inworld STT-1 after 10.60%, at the median.

Test your voice agent the same way.

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