September 23, 2026
Voice Bot Audio Arrives Sooner: Neuro TTS Streams It Within 120 ms
Neuro TTS became generally available on September 22 and starts streaming audio 120 ms after a request. The neuro_tts_v2 version works over REST and WebSocket, returns WAV or PCM at three sample rates, and offers 16 voices. The API plan starts at 2,000 ₽ per month for 1,000 minutes of synthesis and 5 concurrent requests.
Previously, a voice response arrived as a completed file after synthesis. Now Neuro TTS starts streaming audio in chunks: the API requires a POST request with a key, text, and `model_id=neuro_tts_v2`.
Entry price. The free plan provides 50 minutes in Playground, but does not include the API. After the package, each minute costs 2 ₽, while an additional concurrent request costs 1,000 ₽ per month.
For closed networks, Neuro.net says it can deliver the same models in containers without internet access.
Source
