AssemblyAI STT 3.7 Realtime Agent vs Murf Finch-Global

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
  • Finch-Global3.17% · 51ms

AssemblyAI STT 3.7 Realtime Agent is ahead of Murf Finch-Global 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 AgentFinch-Global
  • Overall WER2.41%3.33%
  • Pipecat Dataset WER1.68%3.17%
  • Ocular Dataset WER3.81%3.62%
  • Time to first text493ms1.83s
  • Final-text delay47ms51ms

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 AgentFinch-Global
  • ELO rating10241001–1039 · rank 1–16998981–1017 · rank 5–21
  • Expected win rate54%46%
  • Ocular turns WER · 166 draws of 19818 won14 won
  • Pipecat WER · 807 draws of 999152 won40 won
  • Semantic WER · 193 draws of 1984 won1 won
  • Phone line · 164 draws of 19822 won12 won
  • Call-center noise · 130 draws of 19737 won30 won
  • Babble noise · 117 draws of 19753 won27 won
  • TV noise · 144 draws of 19834 won20 won
  • Reverb · 123 draws of 19762 won12 won
  • Short replies · 63 draws of 705 won2 won
  • Entities · 41 draws of 507 won2 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 Finch-Global 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 Murf Finch-Global 3.33%, so AssemblyAI STT 3.7 Realtime Agent makes fewer errors by 0.91 points. On the 300 licensed Ocular turns, AssemblyAI STT 3.7 Realtime Agent scores 1.68% and Murf Finch-Global 3.17%, so AssemblyAI STT 3.7 Realtime Agent makes fewer errors by 1.49 points. Word error rate counts incorrect, missing, and extra words, so lower is better.

  • Yes. AssemblyAI STT 3.7 Realtime Agent finishes the transcript 1.34s sooner. AssemblyAI STT 3.7 Realtime Agent has the final transcript 493ms after the speaker stops and Murf Finch-Global 1.83s, 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 1.34s 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 Murf Finch-Global after 3.62%, at the median.

Test your voice agent the same way.

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