GPT Realtime Whisper vs Speechmatics Linden
GPT Realtime Whisper is ahead on the Pipecat Dataset and the Ocular Dataset, and Speechmatics Linden on time to first text and final-text delay.
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.
- GPT Realtime Whisper
- 2.16%4th
- Speechmatics Linden
- 2.46%6th
GPT Realtime Whisper, 0.30 points lower
Word error rate on eight licensed recordings provided by Ocular.
- GPT Realtime Whisper
- 3.53%4th
- Speechmatics Linden
- 3.93%6th
GPT Realtime Whisper, 0.41 points lower
Median delay before the first words arrive, from the first audio packet.
- GPT Realtime Whisper
- 1.38s8th
- Speechmatics Linden
- 1.19s7th
Speechmatics Linden, 189ms sooner
Median delay after the speaker stops before the transcript is final.
- GPT Realtime Whisper
- 543ms12th
- Speechmatics Linden
- 147ms6th
Speechmatics Linden, 396ms 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 | GPT Realtime Whisper | 1.38s | 1.84s | 2.86s |
| Speechmatics Linden | 1.19s | 1.56s | 2.52s | |
| Final-text delay | GPT Realtime Whisper | 543ms | 647ms | 669ms |
| Speechmatics Linden | 147ms | 192ms | 213ms |
Both against every other model
Open on the metric where the two sit furthest apart, with the other three a tab away.
Final-text delay after the speaker stops. Lower is better.
| # | Model | Final text | turns | |
|---|---|---|---|---|
| 1 | Inworld STT-1 | 41ms | 206 | |
| 2 | Deepgram Nova-3 | 103ms | 202 | |
| 3 | Deepgram Flux English | 108ms | 199 | |
| 4 | Deepgram Flux Multilingual | 114ms | 175 | |
| 5 | Cartesia Ink 2 | 123ms | 198 | |
| 6 | Speechmatics Linden | 147ms | 194 | |
| 7 | AssemblyAI Universal 3.5 Pro | 180ms | 205 | |
| 8 | Smallest Pulse | 201ms | 206 | |
| 9 | Gradium | 212ms | 198 | |
| 10 | Reson8 | 235ms | 201 | |
| 11 | Google Chirp 3 | 475ms | 206 | |
| 12 | GPT Realtime Whisper | 543ms | 206 | |
| 13 | Google Chirp 2 | 568ms | 205 | |
| 14 | GPT-4o Mini Transcribe | 630ms | 206 | |
| 15 | GPT-4o Transcribe | 652ms | 206 |
Median across the same 206 Ocular Dataset turns. Providers finalize differently, so compare models with similar finalization contracts before treating small gaps as meaningful.
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 |
|---|---|---|---|---|
| GPT Realtime Whisper | 1,000 | 8 | 206 | 0 |
| Speechmatics Linden | 998 | 8 | 194 | 12 |
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 GPT Realtime Whisper or Speechmatics Linden more accurate?
On the 1,000 public Pipecat clips, GPT Realtime Whisper scores 2.16% and Speechmatics Linden 2.46%, so GPT Realtime Whisper makes fewer errors by 0.30 points. On the eight licensed Ocular recordings, GPT Realtime Whisper scores 3.53% and Speechmatics Linden 3.93%, so GPT Realtime Whisper makes fewer errors by 0.41 points. Word error rate counts incorrect, missing, and extra words against the reference transcript, so lower is better.
Which is faster, GPT Realtime Whisper or Speechmatics Linden?
Speechmatics Linden returns first text 189ms sooner, at 1.19s against 1.38s at the median. Speechmatics Linden has the final transcript 396ms sooner, 147ms after the speaker stops against 543ms. Both are measured on the same 206 streamed turns.
How were GPT Realtime Whisper and Speechmatics Linden 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. GPT Realtime Whisper returned usable text for 1,000 of the public clips and Speechmatics Linden for 998. Latency uses each provider's own supported finalization contract.