Soniox · read on October 27, 2026
Soniox TTS Real-Time v2
Soniox TTS Real-Time v2 is a TTS (text-to-speech) model from Soniox. Coval measures its P95 time to first audio at 324.7 ms, which puts it in the Visible delay speed tier. It ranks 15 of 86 in the full ranking.
- Speed tier
- Visible delay
- P95 / median
- 324.7 ms / 264.4 ms
- Listener Elo
- 1,180
- Price per 1M chars
- $13 to $14.81
Where it ranks
Full ranking: 15 of 86. Use case ranks, by speed:
The tier and the ranks are computed by this site from the benchmarks' figures, not declared by Soniox.
Measurements and sources
Coval
| Median time to first audio (P50) | 264.4 ms |
| Time to first audio, P95 | 324.7 ms |
| Time to first audio, mean | 260.3 ms |
| Time to first audio, P90 | 311.3 ms |
| Timed requests | 2,365 |
Speko
| Naturalness score | 1,606 |
| Latency, board chart | 134 ms |
| Latency, model page sweep (median) | 381 ms |
| Numbers and dates pronunciation | 0.49 |
Artificial Analysis
| Listener Elo | 1,180 |
| Listener Elo, 95% interval | 11 |
| Listener Elo, customer service | 1,213 |
| Hard passages pronounced correctly | 78.8% |
Cekura
| Time to first audio, streamed text | 183 ms |
| Time to first audio, text sent whole | 241 ms |
| Reading accuracy | 83.6% |
| Estimated naturalness (1 to 5) | 4.37 |
Details
| Provider | Soniox |
| Access | Proprietary |
| Streaming | A benchmark measures its time to first audio |
| Features listed by Coval | multilingual, voice-cloning |
| Languages tested | Spanish, German, French, Japanese, Chinese, Arabic, Hindi, Filipino, Norwegian, Thai |
| Measured by | Artificial Analysis, Cekura, Coval, Speko |
| Official site | soniox.com |
Sources
- https://artificialanalysis.ai/text-to-speech (read on October 27, 2026)
- https://artificialanalysis.ai/text-to-speech/leaderboard/provider-voice (read on October 27, 2026)
- https://benchmarks.cekura.ai/text-to-speech (read on October 27, 2026)
- https://benchmarks-api-6wxgp27p2a-ue.a.run.app/v1/results/aggregates?benchmark=TTS&window=30d&include_series=false (read on October 27, 2026)
- https://benchmarks.speko.ai/tts/ (read on October 27, 2026)
Values are copied from these pages on the date shown. The match between this model and its benchmark entries is inferred, because a source writes the name differently: Cekura: “Soniox TTS RT v2”; Coval: “Soniox TTS Rt v2”; Speko: “Soniox tts-rt-v2”. See the methodology.
Compare it with the reference model: Gradium TTS vs Soniox TTS Real-Time v2.