Deepgram Nova-3 vs Inworld STT-1
Inworld STT-1 is ahead on the Pipecat Dataset and final-text delay, and Deepgram Nova-3 on the Ocular Dataset. The two tie on time to first text.
Quality and speed frontier
Every model's error rate on the public clips, against how long its transcript takes to settle.
The four metrics, side by side
Word error rate on 1,000 clips from Pipecat's STT benchmark dataset.
- Deepgram Nova-3
- 3.46%10th
- Inworld STT-1
- 2.72%8th
Inworld STT-1, 0.75 points lower
Word error rate on eight licensed recordings provided by Ocular.
- Deepgram Nova-3
- 4.39%10th
- Inworld STT-1
- 10.60%14th
Deepgram Nova-3, 6.21 points lower
Median delay before the first words arrive, from the first audio packet.
- Deepgram Nova-3
- 1.06s6th
- Inworld STT-1
- 1.06s5th
Level
Median delay after the speaker stops before the transcript is final.
- Deepgram Nova-3
- 103ms2nd
- Inworld STT-1
- 41ms1st
Inworld STT-1, 62ms sooner
Lower is better on every metric. The small figure is each model's place in the whole field.
First text and final text, at three percentiles
Both measured on the same 206 streamed Ocular Dataset turns. P90 and P95 say what a caller meets on a bad turn, which a median hides.
| Measure | Model | P50 | P90 | P95 |
|---|---|---|---|---|
| Time to first text | Deepgram Nova-3 | 1.06s | 1.10s | 2.07s |
| Inworld STT-1 | 1.06s | 1.07s | 1.07s | |
| Final-text delay | Deepgram Nova-3 | 103ms | 143ms | 155ms |
| Inworld STT-1 | 41ms | 60ms | 65ms |
Both against every other model
Open on the metric where the two sit furthest apart, with the other three a tab away.
Word error rate on eight licensed recordings provided by Ocular. Lower is better.
| # | Model | Ocular Dataset | clips | |
|---|---|---|---|---|
| 1 | Reson8 | 2.93% | 8 | |
| 2 | Cartesia Ink 2 | 3.09% | 8 | |
| 3 | Smallest Pulse | 3.46% | 8 | |
| 4 | GPT Realtime Whisper | 3.53% | 8 | |
| 5 | Google Chirp 2 | 3.80% | 8 | |
| 6 | Speechmatics Linden | 3.93% | 8 | |
| 7 | Deepgram Flux English | 4.18% | 8 | |
| 8 | Google Chirp 3 | 4.19% | 8 | |
| 9 | AssemblyAI Universal 3.5 Pro | 4.23% | 8 | |
| 10 | Deepgram Nova-3 | 4.39% | 8 | |
| 11 | Gradium | 4.45% | 8 | |
| 12 | GPT-4o Mini Transcribe | 4.45% | 8 | |
| 13 | Deepgram Flux Multilingual | 4.53% | 8 | |
| 14 | Inworld STT-1 | 10.60% | 8 | |
| 15 | GPT-4o Transcribe | 12.45% | 8 |
Counts are the clips each model returned usable text for, so they differ between models and between datasets.
Both on the other two views
Pipecat vs Ocular accuracy
Every model's error rate on the public clips, against the same model on real conversations.
Transcription response time
How soon each model returns its first words, against how soon its transcript is final.
What the numbers rest on
| Model | Pipecat clips | Ocular recordings | Turns timed | Failed |
|---|---|---|---|---|
| Deepgram Nova-3 | 1,000 | 8 | 202 | 0 |
| Inworld STT-1 | 1,000 | 8 | 206 | 0 |
A clip counts only where the model returned usable text, so counts differ between models. Word error rate is measured the same way throughout: incorrect, missing, and extra words against the reference transcript.
Frequently asked questions
Is Deepgram Nova-3 or Inworld STT-1 more accurate?
On the 1,000 public Pipecat clips, Deepgram Nova-3 scores 3.46% and Inworld STT-1 2.72%, so Inworld STT-1 makes fewer errors by 0.75 points. On the eight licensed Ocular recordings, Deepgram Nova-3 scores 4.39% and Inworld STT-1 10.60%, so Deepgram Nova-3 makes fewer errors by 6.21 points. The two datasets disagree, so the better choice depends on whether your audio looks more like clean public clips or real conversations.
Which is faster, Deepgram Nova-3 or Inworld STT-1?
Both return first text after 1.06s at the median. Inworld STT-1 has the final transcript 62ms sooner, 41ms after the speaker stops against 103ms. Both are measured on the same 206 streamed turns.
How were Deepgram Nova-3 and Inworld STT-1 compared?
Both ran the same two datasets under the same conditions as every other model in the benchmark: 1,000 public clips from Pipecat's STT benchmark dataset and eight licensed conversation recordings from Ocular. Deepgram Nova-3 returned usable text for 1,000 of the public clips and Inworld STT-1 for 1,000. Latency uses each provider's own supported finalization contract.