Inworld Realtime TTS-2 Inworld · text-to-speech (TTS) API

Real-time latency

On Coval's 30-day window ending October 1, 2026 (11,657 runs), Inworld Realtime TTS-2 has a P50 (median) time-to-first-audio (TTFA) of 163.1 ms, 7th lowest of the 31 models Coval measures. Its P99, the slowest 1% of turns, is 293.7 ms (6th lowest of the 31 models Coval measures). Latency consistency: a standard deviation of 567.8 ms (29th lowest of the 31 models Coval measures). Mean time to first byte (TTFB) is 123.1 ms. Speko measures a P50 TTFA of 116 ms from US East (n=30), dated July 3, 2026: 9th lowest of the 29 models Speko measures.

Accuracy (word error rate)

Coval measures an accuracy (word error rate) of 4.5% over 11,637 samples on the same window, 8th lowest of the 31 models Coval measures. Speko rates its pronunciation robustness on numbers, dates and currency at 0.61 on a 0-to-1 scale (6th highest of the 28 models Speko rates).

Voice quality

In the Artificial Analysis Speech Arena (capture of October 1, 2026), Inworld Realtime TTS-2 has an Elo of 1,247 (±17, 95% confidence) from 1,513 blind comparisons, 5th highest of the 92 models Artificial Analysis rates. Speko rates its naturalness at 1,561 Elo from blind A/B votes (16th highest of the 41 models Speko rates).

Price

Artificial Analysis lists Inworld Realtime TTS-2 at $20.8 per 1M characters (38th lowest of the 78 models Artificial Analysis lists a price for). Speko lists a cost of about $25 per 1M characters.

Scoring note

Speko changed its pronunciation robustness between the captures of September 30, 2026 and October 1, 2026 without a new measurement date (0.93, then 0.61): the score uses the average, 0.77. The value shown is the latest.

Languages tested

Tested in Spanish, German, French, Portuguese, Japanese, Chinese, Arabic, Hindi, Vietnamese, Korean, Filipino and Thai by Artificial Analysis or Speko. A language missing here was not tested, which does not mean it is unsupported.