Reducing transcription costs by 60% using AWS AI/ML services

This article discusses a solution developed by the Contra Costa County District Attorney's Office, AWS, and AWS Partner ScaleCapacity to automate the manual transcription process of audiovisual evidence files. The solution aimed to reduce transcription costs, improve accuracy, and shorten turnaround times.

Specifically, the article covers:

• Challenges faced by the District Attorney's Office in manual transcription, such as time-consuming process, error-prone, and costly services
• An overview of the serverless, cloud-based solution built using AWS services like Amazon Transcribe, Lambda, API Gateway, S3, DynamoDB, and others
• Key features of the solution, including identifying low-confidence words, highlighting specific words of interest, speaker identification, timestamps, search/replace, and editing capabilities
• Benefits achieved, including 60% cost reduction, faster turnaround time from days to minutes, and improved accuracy with about 80% accuracy before editing
• Future enhancements planned, like support for non-English transcription and translation to English