OuteTTS

OuteTTS TTS

An LLM-based TTS that runs on CPU, GPU, or even in the browser via llama.cpp and Transformers.js.

Sign up for 5,000 character limit

Wrap your text in SSML tags for precise control:

<speak><prosody rate="slow">Slow speech</prosody></speak>

Tags the selected model understands — click to drop one into your text where it happens:

This model reads plain text, so inline tags are ignored. For tag-based emotion, switch to an expressive model like Orpheus or Bark.

Define custom pronunciations (word = pronunciation):

-12 +12
0.5x 2.0x
Free with Piper, VITS, MeloTTS
Your generated audio will appear here. Choose a model, enter text, and click Generate.
Audio Generated Successfully
0:00
Download Audio Download .srt Link expires in 24h
Free tier: personal use. Commercial license from $5/mo
Love TTS.ai? Tell your friends!

About OuteTTS

OuteTTS by OuteAI takes a language-model approach to speech: it extends an LLM with text-to-speech capability while keeping the original architecture intact, so it can run through standard LLM tooling. That gives it unusually broad backend support — llama.cpp on CPU or GPU, Hugging Face Transformers, ExLlamaV2, VLLM, and even in-browser inference via Transformers.js. It is a natural fit for edge deployment and low-resource environments where running a model client-side or on CPU matters more than raw speed. On TTS.ai it is offered on the free tier for English. Because the LLM-based pipeline is slow on long inputs, it is best used for shorter regular text rather than long-form cloning.

Best for: Edge deployment, browser-based TTS, low-resource environments

Browse all OuteTTS voices

At a glance

Developer
OuteAI
License
Apache 2.0
Tier
free
Speed
slow
Voice cloning
No
Languages
English
Max characters
1000

OuteTTS voices

Female 1 (Neutral)

English
Free Female

OuteTTS TTS — FAQ

Across many backends — llama.cpp (CPU or GPU), Hugging Face Transformers, ExLlamaV2, VLLM, and even directly in the browser through Transformers.js — because it preserves the underlying LLM architecture.

Its LLM-based design runs efficiently on CPU and in the browser, so it can operate client-side or on modest hardware without a dedicated GPU.

It works best on shorter inputs. The LLM-based pipeline is slow on long passages, so it is offered for regular short-to-medium TTS rather than long-form generation.
← All voices