AssemblyAI STT 3.7 Realtime Agent vs NVIDIA Nemotron 3.5 Transcribe

Compare

Accuracy and speed

Left gets fewer words wrong. Lower has the final text ready sooner. Nemotron 3.5 Transcribe doesn't support streaming, so it is not shown.

Accurate and fast0ms200ms400ms600ms800ms0%2%4%6%Words wrong on clean audio (Pipecat WER)
  • STT 3.7 Realtime Agent1.68% · 47ms
  • Nemotron 3.5 Transcribe1.89%

NVIDIA Nemotron 3.5 Transcribe is ahead of AssemblyAI STT 3.7 Realtime Agent on both datasets together. 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 AgentNemotron 3.5 Transcribe
  • Overall WER2.41%2.28%
  • Pipecat Dataset WER1.68%1.89%
  • Ocular Dataset WER3.81%3.03%
  • Time to first text493msn/a
  • Final-text delay47msn/a

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

STT 3.7 Realtime Agent vs Nemotron 3.5 Transcribe FAQ

  • NVIDIA Nemotron 3.5 Transcribe is more accurate on the public clips. On the 1,000 public Pipecat clips, AssemblyAI STT 3.7 Realtime Agent scores 2.41% and NVIDIA Nemotron 3.5 Transcribe 2.28%, so NVIDIA Nemotron 3.5 Transcribe makes fewer errors by 0.13 points. On the 300 licensed Ocular turns, AssemblyAI STT 3.7 Realtime Agent scores 1.68% and NVIDIA Nemotron 3.5 Transcribe 1.89%, with likely ranges of 1.47 to 1.90% and 1.63 to 2.17%. The ranges overlap, so the difference is too close to call. Word error rate counts incorrect, missing, and extra words, so lower is better.

  • They cannot be compared on speed. NVIDIA Nemotron 3.5 Transcribe was tested on whole recordings rather than a live stream, so its speed is not measured.

  • The two are too close to call on real conversations. NVIDIA Nemotron 3.5 Transcribe was tested on whole recordings rather than a live stream, so its speed is not measured. A voice agent cannot reply until the final transcript arrives, and the Ocular recordings are real conversations.

  • NVIDIA Nemotron 3.5 Transcribe was tested on whole recordings rather than a live stream, so whether it streams partial text was not measured. AssemblyAI STT 3.7 Realtime Agent returns partial text before the speaker stops.

Test your voice agent the same way.

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