Our continuous search to deliver a better worldwide viewer experience

Aware of the globalized world in which we live in, VoiceInteraction is constantly committed to widening the series of its linguistic models. As such, VoiceInteraction entered 2020 focused on continuing the work of expanding the range of languages offered and dedicating itself not only to this development but also to the constant improvement of the languages ​​already included in the portfolio.

The work developed by VoiceInteraction during the year of 2019 ensured improvements in the accuracy of the language models from 7% and up to 60%.

Regarding the existing languages​ in the VoiceInteraction’s portfolio, the company’s work that was carried out led to improvements in the models, namely between 6.6% and 19.3%, in relative terms. If we focus on the new languages, ​​added to VoiceInteraction ’s portfolio in 2019 (namely Dutch, the English variety from Canada and India, and the Spanish variety from the United States), we see improvements between 15.4% and 50.6%, regarding the models’ accuracy and when comparing the first linguistic model developed and the most recent ones.

At the same time, VoiceInteraction also focused on improving the Portuguese speaking models (Portugal and Brazil) as well as the models dedicated to the transcription of town hall meetings (with a focus on the linguistic models of Portuguese and the Irish variety of English).

For 2020, VoiceInteraction plans on adding new languages ​​to its portfolio and is currently working on acquiring data from languages such as Hindi, Catalan, Russian, Turkish, Arabic, and Hebrew, in order to offer a better viewer experience, one that is more global and integrated to its customers.