Evaluate video & audio of mock personal interviews of youth, and thematic videos made by children on varied topics.
-Extraction of features from these video/audio files.
-Development of a framework to evaluate themes & concepts from the video files.
In Pratham’s Hybrid Learning communities, it has been found that with the right tools, learners can use their time to build creative skills while sharing their stories, opinions, and ideas. Once they make these videos, having a portal on which these videos can be uploaded to obtain feedback, enables scope for creativity. Having a real audience shows learners that their efforts can matter, such as having an app to get feedback on how they are doing and how they can improve. The basic aim of the feature extraction app will be to analyze language and vocabulary, and detect faces, objects, texts (if any) in the video submitted by the learner and then come up with a report of how relevant the video was with the expected outcomes of the activity.
-The video/audio files are in English, Hindi, Marathi & Kannada.
-Video & audio assigned to volunteers will be in the language of their preference.
-Volunteers will be provided with the video/audio & the initial transcription that we obtained from Google video intelligence tool for videos and Google speech to text (STT) API for audios.
-The transcript obtained from Google speech recognition does not work well for Indian languages. Volunteers are required to update this transcript to exactly what is being spoken in these videos/audios.
-Volunteers are encouraged to progress to the next audio/video file as per interest and availability of time.
For this activity, volunteers are requested to download the respective language keyboard based on their preferences:
Hindi : http://rajbhasha.net/download2/index.php/GoogleInputToolsHindi.exe
Marathi : http://rajbhasha.net/download2/index.php/GoogleInputToolsMarathi.exe