Unreal Speech

Unreal Speech is a fast, low-cost text-to-speech API for developers. It focuses on price and speed: audio streams back in around 300 milliseconds, and you can generate very long files of up to 10 hours from a single request.

The API offers 48 voices across 8 languages and returns per-word timestamps, so you can highlight words in sync with the audio, which is useful for reading apps, e-learning, accessibility tools and video subtitles. It is powered by the open Kokoro-82M speech model, and pricing gets cheaper per character as your volume grows. A web studio lets non-developers create audio without code.

Good fit if you: need affordable, production-ready text-to-speech at scale, such as audiobooks, articles, apps or voice features.
Less ideal if you: need voice cloning or highly emotional, character-style acting voices.

Pricing: free tier with 250,000 characters to start; paid plans scale with usage and get cheaper at higher volumes.