Neon AI speaks 25 languages, and understands 6 languages ...and that's just the beginning on our platform.
Our efficient, lightweight Speech-To-Text (STT) & Text-To-Speech (TTS) systems use new techniques including transfer learning to enhance our machine-learning models. We keep our on-device polylingual system lightweight by working directly with graphemes, rather than including a phoneme set for each language. With our new methodology, even very small data sets can be enough to produce an excellent result. For example, we trained a high quality English "voice" for Neon in just 8 days, with only 20 hours of recorded speech to learn from. With the transfer learning technique, we were then able to give Neon an Irish speaking voice in 4 days, using only a 10 minute speech sample of Irish.
At Neon AI we've expanded our language support far beyond English, using Coqui and other resources to provide on-device capabilities for maximum privacy and security.
On-device, cloudless translation for every global language is within reach.
Estonian, Irish, Maltese - our innovative AI training techniques make it practical to add even languages with relatively few native speakers. Have a specific language you'd like prioritized? Great! Our commitment to open source means you can find out exactly what we're doing, and support us or use our innovations to support your own projects. Join us on GitHub, submit a subscription form, or email any of us directly to get involved.
Languages currently supported on-device, ordered by number of speakers:
English, Spanish, French, German, Italian, Polish, Ukrainian, Dutch, Romanian, Hungarian, Greek, Czech, Swedish, Portuguese, Croatian, Bulgarian, Danish, Slovak, Finnish, Lithuanian, Slovenian, Latvian, Estonian, Irish, Maltese
Demonstrations of Neon AI interactions
Transcription, translation and speech services - our real-time language services are growing in real-time themselves now. Integrating new STT and TTS engines is straightforward using Neon AI plugins. Neon AI plugins are now available for many cloud services including AWS and Google, and for connecting to on-premises services.
On KlatChat, real-time polylingual translation and AI voice commands are provided for applications such as assembly instructions, retail floor automation and caregiving situations.