520.2 ms P50 time-to-first-audio on Coval; ranked 34th (tied) of 112 in the TTS for Voice Agents 2026 index.
Benchmark summary: 14.4/100, ranked 34th (tied) of 112 in the TTS for Voice Agents 2026 index (Listed). Strongest criterion: accuracy, 7.5 of 15 points.
Deepgram Aura 2 En is a proprietary text-to-speech (TTS) API from Deepgram, served by Cloudflare, measured by one independent benchmark, Coval.
Overall score
14.4 /100
Data checked on September 30, 2026
Updated after each benchmark capture
Every point comes from a public benchmark: sources · methodology
Deepgram Aura 2 En ranks 34th of 112 in the 2026 text-to-speech (TTS) API benchmark, with a score of 14.4/100.
On Coval's 30-day window ending October 1, 2026 (1,116 runs), Deepgram Aura 2 En has a P50 (median) time-to-first-audio (TTFA) of 520.2 ms, 28th lowest of the 31 models Coval measures. Its P99, the slowest 1% of turns, is 1,696.9 ms (28th lowest of the 31 models Coval measures). Latency consistency: a standard deviation of 326.6 ms (24th lowest of the 31 models Coval measures). Mean time to first byte (TTFB) is 360.6 ms.
Coval measures an accuracy (word error rate) of 2.5% over 1,116 samples on the same window, the lowest of the 31 models Coval measures.
Neither Artificial Analysis nor Speko rates its voice quality, so it counts 0 of 20.
No price is published by Artificial Analysis or Speko for this model; price counts 0 of 5.
Coval measures this model through Cloudflare's API, so the latency above is Cloudflare's. Coval's model registry names Deepgram as the creator and Cloudflare as the host.
OpenAI GPT-4o mini TTS is a proprietary text-to-speech (TTS) API from OpenAI, measured by two independent benchmarks: Coval and Speko.
Nari Qwen3-TTS is an open-weights text-to-speech (TTS) API from Alibaba (Qwen), served by Nari Labs, measured by one independent benchmark, Speko.
Google Chirp 3: HD is a proprietary text-to-speech (TTS) API from Google, measured by two independent benchmarks: Artificial Analysis and Coval.