Speech processing technology specialist VoiceInteraction is presenting Audimus.Media in Hall 14 this week, showcasing a software solution that generates subtitles automatically and in real time for live TV shows and live streaming channels.
The solution combines advanced signal processing and speech recognition with vocabulary and segmentation techniques, enabling captioning with a high degree of precision (the company claimed this was over 95 per cent).
Audimus.Media can also identify speakers and detect speaker transitions, increasing the readability and flow of subtitles. For post-production purposes, a subtitle file can also be exported with timecodes.
Different language models (British English, American English, French, German, Spanish, Portuguese, Italian) with specific themes are provided, such as news or sports. The total vocabulary exceeds 200,000 words.
The technology enables the adjustment of the phonetic transcriptions to adapt models to local language.