Unreal Speech
Unreal Speech is a text-to-speech API offering low-latency streaming, per-word timestamps for word-by-word highlighting, short/medium/long synthesis endpoints, a websocket real-time stream, and tiers for large monthly character volumes including a free allowance.
Unreal Speech publishes a free tier plus multiple monthly subscriptions with fixed character allowances and per-month pricing shown on the pricing page. Several paid tiers include explicit $/mo amounts; one plan entry on the linked official page shows a promotional price phrase that is ambiguous.Is Unreal Speech right for you?
Good for
- Product or mobile teams that need streaming TTS with per-word timestamps for in-app highlighting and large-batch synthesis at documented monthly character tiers.
- Provides per-word timestamps and a websocket endpoint to stream audio with timing for word-by-word highlighting.
- Documents low latency streaming and a fast start endpoint suitable for real-time playback (around 0.3s).
Keep in mind
- The FAQ explicitly states voice cloning / custom voice creation is not available today.
- Projects that require built-in custom voice cloning today; the vendor FAQ states custom voice creation is not available yet—confirm roadmap timing if needed.
Pricing and plan limits
The linked official pages list specific monthly prices for Free, Starter, Plus, Pro, and Enterprise plans and show a Basic line with a promotional phrase that makes the exact billed rate ambiguous; Custom requires a high-volume inquiry.
More about capabilities and limits
Documented strengths
Provides per-word timestamps and a websocket endpoint to stream audio with timing for word-by-word highlighting.
Documents low latency streaming and a fast start endpoint suitable for real-time playback (around 0.3s).
Documents multi-size synthesis endpoints including an asynchronous task for very long input (up to 500,000 characters).
Limitations to consider
The FAQ explicitly states voice cloning / custom voice creation is not available today.
Choosing a plan
Run a proof-of-concept using the websocket streamWithTimestamps and your target voice/latency scenarios to validate timestamp accuracy and end-to-end playback latency.
Based on the linked product documentation.
Available on
Api.
Official sources
Alternatives to Unreal Speech
Choose around the work you need to do.
Cartesia
Cartesia supplies streaming speech models and instant voice cloning with cloud, on‑prem, and on‑device deployment options plus managed voice-agent features for production TTS and telephony integration.
Fish Audio
Fish Audio delivers a developer-accessible low-latency TTS and voice-cloning API with a documented free tier, broad language support, and sub-100ms first-audio latency claims for interactive applications.
