To stand out at this year’s NAB Las Vegas,  VoiceInteraction had prepared a range of new features and benefits for those who use or might be interested in using our Automatic Closed Captioning System.

Take a look at some of those new highlights:

  • Optimization of our Automatic Speech Recognition (ASR) engine with a better memory management and an efficient use of new threads in parallel that improves speed and decreases latency;
  • With the daily collection of data, both from text and acoustic, the training of the models has been producing constant and consistent improvements. With the use of more GPUs in parallel, we were able to decrease the time needed to train these models. This allows us to be currently generating daily models for several languages, ensuring that our customers have access to updated models every day;
  • We have conducted some training on the Speaker Turn, Speaker ID and Punctuation models that will allow us to present better results;
  • Substantial improvement in several languages: American English, American Spanish, Brazilian Portuguese, Spanish from LATAM countries (Argentina, Chile, Peru, Colombia, generic Caribbean) – with specific models for each country; Mexico, European languages (Italy, France, Spain, Germany, Holland, Portugal, Switzerland). We also started the development of models for Asia (Asian English, Indian English, Hindu, Malay and Korean);
  • It is now possible to work with 2 or more languages at the same time (for example, English and Spanish in the same program);
  • Use of Third-party technology for real-time translation: English -> Spanish and Spanish -> English, and the possibility to send simultaneously as CC1 and CC3 in 2 languages;
  • Integration with AWS and Google – cloud – with the possibility to do other integrations on-premises;
  • Integration with Text-to-Speech systems, such as Acapela, AWS cloud and Google cloud, among others;
  • New functionalities in the scheduling interfaces with the addition of new words in the vocabulary, integration with MOS / ENPS, iNews and Dalet. These integrations will allow access to more data and a more effective update of the daily models;
  • Control of the application by GPIO, which is an alternative to the interface and mute/unmute or use of the REST API;
  • Integration with several cards, such as Decklink, Avermedia (with transcode) and AJA;
  • Support for different stream formats by adding WebVTT, Unified Streaming, YouTube, Facebook or Zoom;
  • Export of video and CC, whether in compressed or non-compressed formats, to use on VOD platforms.

VoiceInteraction remains focused on the future and is working daily to deliver a more solid product base to its customers.

If you are interested or have any questions about our product, do not hesitate and contact us: info@voiceinteraction.pt