We Support the Easier Watching of Doğuş Media Group Content With Our Google Cloud Partnership!

Doğuş Media Group has been one of the leading media companies in Turkey since 1999, providing communication channels on television, radio, and internet. The remarkable number of viewers resulting from the diversity of the programs and the channel line ups, has always made Doğuş Media Group's technological requirements unavoidable. The focus on reaching the end customer promptly has been one of the broadcasting principles of Doğuş Media Group, which has added the notion of “becoming a worldwide broadcasting company that touches people and employs innovative technology” to its mission. Therefore, while the balance between technology and audience is the primary source of technological requirements, the principles that should not be compromised have also guided the solution to the problem.

Increase in Technological Capacity Thanks to Google Cloud Platform!

The respective supervisory and regulatory agency has made the usage of closed captions mandatory for broadcasters. Doğuş Media Group then attempted to identify the fastest and best option for adding subtitles to internet and television content for individuals who wish to watch these contents at a low volume or have hearing impairments. In order to get the desired result, both cutting-edge technology and convenience of use were considered for the quality of subtitles, rather than the preference for traditional methods. As a modern solution, a system called “Dsub” that offers a human touch was proposed in the final stage.

In the process of adding subtitles to videos, D-sub used Python for data processing, then GCP ASR (Google Cloud Platform Automatic Speech Recognition) API (Application Programming Interface) for automatic speech-to-text translation, and finally BigQuery to store the results of the ASR API. The results collected from the broadcasting agency were adjusted and provided to Google Data Studio to allow the visualization process through further analysis-modelling and self-supervised learning.

A More Competent Business Solution with Self-Learning Method and D-Sub

The error rate in subtitles has been minimized by using the GCP ASR API, and the performance has resulted in encouraging academic results and a commercially feasible system. Subtitle translation time was reduced, and the end user was immediately reached thanks to the first automatic speech recognition (ASR) system that used cloud services. Individuals with a hearing impairment have thus achieved a complete content viewing experience. Furthermore, labour and time were saved thanks to the self-learning method, and our colleagues were able to focus more on the operation of the control and inspection mechanism of the developed product.