Finding out the best engine for transcription (ASR), Translation (MT) and Voice-Over (VO). That is what the DW Benchmarking tool is about. Once the best engine or model is identified, it can be used for improving user interactions (providing shortcuts) on available platforms (e.g. in plain X).
Image shows three areas: ASR, MT and VO. Buttons in each area lead to results and to further options, including manual and automatic evaluation.
SELMA contribution
- Development of a web-based UI
- Integration of further models
- Collection of manual and automatic evaluation
More
- SELMA Benchmarking website (credentials needed): https://selma-benchmarking-frontend.vercel.app/
Bring me to Diversity Monitoring Prototype