Deepgram Aura 2 Deepgram · text-to-speech (TTS) API

Speed

On Coval's 30-day window ending October 22, 2026 (3,308 runs from US East), Deepgram Aura 2 starts speaking after 266.4 ms on a typical turn (median) and 526.2 ms on a slow turn (P95, the slowest one in twenty). That P95 is 23rd lowest of the 32 models Coval measures, which puts it in the Visible delay tier: the voice becomes a visible part of the delay. On the median it is a tie (gap under 10 ms) with Rime Mist v3, Soniox TTS Real-Time v2 and Soniox TTS RT v1. Speko's board plots it at 147 ms time to first audio from US East; its model page reports a 125 ms sweep.

Voice quality

Speko rates its naturalness at 1,584 Elo from blind A/B votes on phone-agent lines (9th highest of the 43 models Speko rates).

Numbers and dates

Speko's pronunciation test on numbers, dates and currency scores it 0.2 on a 0-to-1 scale (25th highest of the 30 models Speko scores). This is the accuracy measure used here; Coval's word error rate separates models too little to compare them.

Price

Coval lists $30 per 1M characters (since Aug 10, 2026). Speko lists about $30. List prices, as each source prints them.

Languages tested

Tested in Spanish, German and French by Artificial Analysis or Speko. A language missing here was not tested, which does not mean it is unsupported.