Luna:Text to Speech Demo

Cobalt's Luna technology powers all our text-to-speech projects. For this limited demo, spell out all the text as you want it said: e.g. "nineteen eighty four" or "one thousand nine hundred eighty four" rather than "1984". Why?

For a production deployment of Luna, we specify custom rules to determine how numbers, symbols, acronyms, etc. should be verbalized. We do not include these tokenization rules for the demo because the same input may be verbalized differently in different contexts (e.g. "7-11" might be "the seventh of November" in the U.K. or "July eleventh" in the U.S. or even "Seven Eleven" the brand name).

{{ currentIconPlay }} Play cloud_download Download

This page is provided for demonstration purposes only and has been restricted accordingly. Commercial use of the generated speech is not allowed. The text you upload may be stored on our servers for internal purposes.


State of the art

Luna uses deep-learning to build sophisticated, best in class models. Our parametric synthesis models produce speech that sounds natural, highly intelligible and customizable.


Embedded / Cloudbased

Luna can run on your device or behind your firewall so that your sensitive data is protected. It can also run as a secure cloud service.



We offer Luna -- the runtime as well as custom-voice services at very competitive rates. Talk to us to find out more!



Most cloud-based TTS services allow you to choose from a prebuilt set of voices. We can work with you to build an exclusive, custom voice for your product/brand.


Streaming synthesis

Luna offers a streaming mode: instead of waiting for the entire utterance to be synthesized, Luna can send audio buffers as they are generated, for a fluid, low-latency user experience.


Supported Languages

Luna supports a variety of languages such as English and Japanese. We can also build a custom voice for you in a language you need!