Based strenuous efforts towards innovations, GTCOM launched a new machine translation service model based on hardware-software server. By embedding AI algorithms and engines for image recognition, video analysis, NPL, knowledge graph and more, it provides all-in-one machine translation solutions. Specifically, there are the webpage translation plug-in, the document translation system, YeeCaption and video analysis software for the translation of captions and subtitles, and PhotoTrans for images. Hardware-software integration connects them all seamlessly and enables private deployment. Thus, resource allocation can be more reasonable, translation can be done more efficiently, and the system runs with higher stability, meeting customers’ needs in all aspects while ensuring security. Upgrading and maintenance for both the hardware and software can be provided on a quarterly basis.
In “accurate” model, the speed is up to 800 characters per second per card, while in “concise” model, the speed can be up to 2,000 characters per second per card. The G4 concurrency is up to 8,000 characters, or 400 sentences per second, and 28.8 million characters per hour. The G8 concurrency is up to 16,000 characters, or 800 sentences, per second, and 57.6 million characters per hour. This means what this system processes within one hour is equal to what a 140 human translators can do in 100 days. Concurrency: 240 characters/min (G4 concurrency equals 2,000 persons/s and G8 concurrency equals 4,000 persons/s).
For more information:
2020nlp@gtcom.com.cn