AWS’ transcription platform is now powered by generative AI – The Verge

AWS has added new languages ​​to its Amazon Transcribe product, offering generative AI-based transcription for 100 languages ​​and a number of new AI features for customers.

During the AWS re: Invent event, it was announced that Amazon Transcribe can now recognize more spoken languages ​​and start call transcription. AWS customers use Transcribe to add speech-to-text capabilities to their apps in the AWS Cloud.

The company said in a blog post that Transcribe trained on “millions of hours of unlabeled audio data from over 100 languages” and uses self-supervised algorithms to learn patterns of human speech in different languages ​​and accents. AWS ensured that some languages ​​were not overrepresented in the training data to ensure that less commonly used languages ​​were as accurate as more commonly spoken languages.

At the end of 2022, Amazon Transcribe supported 79 languages.

According to AWS, Amazon Transcribe has 20 to 50 percent accuracy in many languages. It also offers automatic punctuation, custom vocabulary, automatic speech recognition, and custom vocabulary filters. It can recognize speech in audio and video formats as well as in noisy environments.

The Verge reached out to AWS for information about past accuracy and the base models used for Amazon Transcribe.

AWS says Amazon Transcribe’s advances in better speech recognition are also leading to better accuracy in its call analytics platform, which is widely used by its contact center customers. Amazon Transcribe Call Analytics, now also based on generative AI models, summarizes interactions between an agent and a customer. AWS said this reduces post-call work when creating reports and allows managers to quickly read the information without having to go through the entire log.

Of course, AWS isn’t the only company offering AI-powered transcription services. Otter has been making AI transcriptions available to consumers and businesses for some time and released a summarization tool in June. While not quite the same, Meta announced that it is working on a generative AI-powered translation model that recognizes nearly 100 spoken languages.

AWS also announced additional features for its Amazon Personalization product that will allow customers to offer products or show them recommendations, such as how streaming services can suggest new shows based on previous activity. AWS has added Content Generation, which writes titles or email subject lines to connect recommendation lists thematically.