Amazon Web Services (AWS) has added new languages to its Amazon Transcribe service, providing AI-driven generative transcription for 100 languages and a wide array of new AI tools for its customers.
By the end of 2022, AWS customers were using Amazon Transcribe to add transcription capabilities to their AWS cloud-based apps. As announced by Amazon during their AWS re:Invent event, Amazon Transcribe can now recognize a much larger number of spoken languages and transcribe automatically.
The company states in a blog post that Transcribe has been trained on “millions of hours of unlabeled audio data from over 100 languages” and employs self-supervised algorithms to learn patterns of human speech across different languages and accents.
AWS claims to have ensured that some languages were not overrepresented in the training data to ensure that less commonly used languages were as accurate as the more widely spoken ones.
Amazon Transcribe boasts an accuracy range of 20 to 50 percent across many languages, according to AWS. It also offers automatic scoring, customized vocabulary, automatic language identification, and custom vocabulary filters. Furthermore, it can recognize human speech in audio and video formats, even in noisy environments.
With improved language recognition, AWS claims that the advancements in Amazon Transcribe also translate into higher accuracy for its Call Analytics platform, often used by their contact center clients.
Amazon Transcribe Call Analytics, now powered by AI generative models, summarizes interactions between an agent and a customer. According to AWS, this reduces post-call reporting work, allowing administrators to quickly glean information without needing to review the entire transcription.