AWS's transcription platform has moved to generative AI

By: Bohdan Kaminskyi | 28.11.2023, 17:09

Amazon

Amazon Web Services has upgraded its Amazon Transcribe speech recognition and transcription service. It now works based on generative AI models and supports 100 languages

Here's What We Know

According to AWS, millions of hours of unlabelled audio data were used to train Transcribe. Thanks to a self-aware algorithm, the service learns the features of human speech in different languages and dialects.

AWS ensures a balanced representation of the data to ensure high accuracy for all languages. The company claims that Transcribe's accuracy varies between 20 per cent and 50 per cent depending on the language.

Transcribe's upgrade has also improved the efficiency of its Call Analytics solution for analysing conversations in contact centres. Managers can now more quickly get summarised call information without listening to recordings.

Source: AWS