Command line app


Sep 2019

Automated speech to text testing tool

This app transcribes and tests audio files with Microsoft, Amazon & Google\'s speech to text services, to help compare accuracy and performance of each service for a given use case.

Project highlights

Speech to text

To transcribe audio, I used Amazon, Microsoft & Google's speech to text services. The app can transcibe and test batches of audio files and store results locally

Analytics & Automated tests

To calculate accuracy, the word error rate (WER) is calculated each transcription is compared against its original transcript. Transcription times for each of the cloud services is recorded to understand performance

Built with

Javascript, Node js, Python, AWS S3, Boto3, Microsoft Speech SDK and Google Speech to Text API

Want to hear more about the project? Get in touch or check my social links