December 2023 Summaries

Speech Intelligence company Speechmatics has added Persian as its 49th supported language for transcription. This addition allows the understanding of an additional 110 million voices, further expanding the range of languages that can be accurately transcribed by AI technology. The process of learning a new language involves gathering diverse audio data and understanding the nuances of written forms. Speechmatics uses self-supervised learning to quickly adapt its models to new languages, achieving high accuracy in transcription.

Speechmatics has launched "Topics," a feature that uses large language models to quickly identify custom topics in content and accurately discover what's being spoken about. This tool can parse transcripts of media, customer conversations, social media videos, podcasts, internal meeting recordings, and more. It offers the ability to analyze transcripts for specific user-defined topics without needing labeled speech data or training a bespoke machine learning model. The Topics API is flexible and returns information such as detected topics, their frequency, timestamps of detection, and related transcript text. Use cases include improving contact center processes, enhancing customer experience, and informing brand workstreams and advertisement targeting in media monitoring. Speechmatics' Topics can also be used in conjunction with other speech capabilities like Sentiment analysis, Custom Dictionary, or Summarization for more comprehensive insights.