We've improved user engagement in conversations with high-quality, natural voices.
Our Text-To-Speech (TTS) solution can be widely used in applications such as virtual assistants, audiobooks, accessibility tools, and automated customer service systems.
In research, the comparative MOS was positive for this model. Also, the real-time factor when used on an RTX3090 is 0.07, one of the fastest on the market, very suitable for voice bots.
Our model retains the intonation and sentiments of the conversation well, such as sadness and happiness. This model can be run on instance or serverless for minimal network delay.
In benchmarks, the system generated audio with higher quality than the original recordings. CMOS +0.28
High speed for use with voice agents. Real-time factor of 0.17
Run on your own instance with high privacy
We offer free UniMRCP drivers. This means you can use native Asterisk and FreeSwitch drivers and can support Cisco and Avaya implementations
We can produce custom voices with a small custom audio set, ideal for low-resource languages
We have REST APIs with examples in Curl, Python and Node
Products