Consumer Goods Company Uses GCP To Turn Videos Into Queryable Database

Team Searce's expertise in leveraging Cloud Speech API, DataStore & Big Query helped design, develop and deploy an intelligent video to text solution in record time.

  • The company is known for its design and varied range of products covering a majority of market segments.
  • They leverage technology to innovate and stand apart from their competitors
The Need to Implement Google Cloud Platform to Create the Right Solutions

The client had hundreds of training videos with thousands of hours of video content over a few years. They wanted to create a solution to search through those videos in real time and find relevant content based on queries.

How Searce used Google Cloud Speech API and BigQuery to streamline the video content

Searce Team collaborated with the company and executed alongside their team to meet the infrastructure and timeline goals:

  • Used Google Cloud Speech API, BigQuery, DataStore, etc to convert the videos to audio and then audio to text and store it in BigQuery
  • Stored data in text chunks along with timestamp associated
  • On querying a particular text, the search results would be video links with highlights in the video seek bar wherever the searched phrase/text was found

There was no ready-made solution available for executing on this project. The key complexity was because of the fact that the client wanted to highlight the part of the video wherever the text was found. It was complicated because of multiple reasons - the number of words in each video was huge and considering that each word should be searchable made the database design very complex. Moreover, the app engine had an upload limitation of 32 MB to Cloud Storage directly. So we had to use resumable upload using signed URLs, upload from browser by XMLHttpRequest requests - Chunk Uploading

A reduction in the total number of questions coming to Knowledge Management team Help for the Knowledge Management team to avoid creating duplicate training materials