AssemblyAI STT 3.7 Realtime Agent vs Deepgram Flux Multilingual

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
  • Flux Multilingual4.96% · 114ms

AssemblyAI STT 3.7 Realtime Agent is ahead of Deepgram Flux Multilingual 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 AgentFlux Multilingual
  • Overall WER2.41%4.81%
  • Pipecat Dataset WER1.68%4.96%
  • Ocular Dataset WER3.81%4.53%
  • Time to first text493ms793ms
  • Final-text delay47ms114ms

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 AgentFlux Multilingual
  • ELO rating10241001–1039 · rank 1–16936916–966 · rank 18–23
  • Expected win rate62%38%
  • Ocular turns WER · 146 draws of 19636 won14 won
  • Pipecat WER · 731 draws of 999240 won28 won
  • Semantic WER · 184 draws of 19610 won2 won
  • Phone line · 148 draws of 19840 won10 won
  • Call-center noise · 101 draws of 19479 won14 won
  • Babble noise · 64 draws of 184103 won17 won
  • TV noise · 119 draws of 19567 won9 won
  • Reverb · 119 draws of 19866 won13 won
  • Short replies · 51 draws of 6914 won4 won
  • Entities · 40 draws of 497 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 Flux Multilingual 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 Deepgram Flux Multilingual 4.81%, so AssemblyAI STT 3.7 Realtime Agent makes fewer errors by 2.40 points. On the 300 licensed Ocular turns, AssemblyAI STT 3.7 Realtime Agent scores 1.68% and Deepgram Flux Multilingual 4.96%, so AssemblyAI STT 3.7 Realtime Agent makes fewer errors by 3.28 points. Word error rate counts incorrect, missing, and extra words, so lower is better.

  • Yes. AssemblyAI STT 3.7 Realtime Agent finishes the transcript 300ms sooner. AssemblyAI STT 3.7 Realtime Agent has the final transcript 493ms after the speaker stops and Deepgram Flux Multilingual 793ms, 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 300ms 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 Deepgram Flux Multilingual after 4.53%, at the median.

Test your voice agent the same way.

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