AssemblyAI vs Deepgram speech-to-text
AssemblyAI is ahead of Deepgram on the Pipecat Dataset, time to first text and final-text delay. The Ocular Dataset is too close to call.
Each row compares the best AssemblyAI model with the best Deepgram model on that metric, from the same tests.
Last updated Methodology by Dileep Chagam
AssemblyAI and Deepgram, metric by metric
| Metric | AssemblyAI | Deepgram |
|---|---|---|
| Pipecat Dataset WER | 1.77%AssemblyAI Universal 3.6 Pro | 3.46%Deepgram Nova-3 |
| Ocular Dataset WER | 4.23%AssemblyAI Universal 3.5 Pro | 4.18%Deepgram Flux English |
| Time to first text | 485msAssemblyAI Universal 3.6 Pro | 791msDeepgram Flux English |
| Final-text delay | 91msAssemblyAI Universal 3.6 Pro | 103msDeepgram Nova-3 |
- WER · Word error rate
- Share of words the model got wrong: replaced, missing or added. Lower is better.
- Pipecat Dataset
- 1,000 short public clips from Pipecat's STT benchmark dataset.
- Ocular Dataset
- 8 recordings from 4 real phone conversations, licensed from Ocular.
- TTFT · Time to first text
- How long after the audio starts streaming until the first words appear.
- TTFS · Final-text delay
- How long after the caller stops until the full transcript arrives. A voice agent waits for this before it replies.
AssemblyAI has 2 models on Converse-STT: AssemblyAI Universal 3.5 Pro and AssemblyAI Universal 3.6 Pro. Its best on accuracy is AssemblyAI Universal 3.6 Pro (1.77%). Its fastest is AssemblyAI Universal 3.6 Pro (91ms).
Deepgram has 3 models on Converse-STT: Deepgram Flux English, Deepgram Flux Multilingual and Deepgram Nova-3. Its best on accuracy is Deepgram Nova-3 (3.46%). Its fastest is Deepgram Nova-3 (103ms).
Which to pick
AssemblyAI is the better pick over Deepgram on every measure that separates them: the Pipecat Dataset (1.69 points lower), time to first text (306ms sooner) and final-text delay (12ms sooner). For the Ocular Dataset, the gap is inside the margin of error, so neither is ahead.
Every model-to-model comparison
- AssemblyAI Universal 3.5 Pro vs Deepgram Flux English
- AssemblyAI Universal 3.5 Pro vs Deepgram Flux Multilingual
- AssemblyAI Universal 3.5 Pro vs Deepgram Nova-3
- AssemblyAI Universal 3.6 Pro vs Deepgram Flux English
- AssemblyAI Universal 3.6 Pro vs Deepgram Flux Multilingual
- AssemblyAI Universal 3.6 Pro vs Deepgram Nova-3
Frequently asked questions
Is AssemblyAI or Deepgram better for speech-to-text?
AssemblyAI is ahead, 1.69 points lower. The best AssemblyAI result is AssemblyAI Universal 3.6 Pro at 1.77%, and the best Deepgram result is Deepgram Nova-3 at 3.46%, on the Pipecat Dataset.
Which is faster, AssemblyAI or Deepgram?
AssemblyAI is faster, 12ms sooner. The best AssemblyAI result is AssemblyAI Universal 3.6 Pro at 91ms, and the best Deepgram result is Deepgram Nova-3 at 103ms, on final-text delay.
Which AssemblyAI and Deepgram models were tested?
AssemblyAI: AssemblyAI Universal 3.5 Pro, AssemblyAI Universal 3.6 Pro. Deepgram: Deepgram Flux English, Deepgram Flux Multilingual, Deepgram Nova-3. Every model ran the same tests on Converse-STT.