August 2022 Summaries

The latest addition to Speechmatics engine allows customers to automatically identify the predominant language spoken in any media file, saving time and effort on manual review, and providing accurate transcription with a confidence score. This update drives efficiency by removing the need for manual selection of language packs, and is particularly useful for media and broadcast organizations with extensive archives of audio content. The Speechmatics engine has been trained using hundreds of thousands of individual voices and millions of hours of unlabelled voice data, resulting in a more accurate and inclusive speech-to-text experience. This update can be used with up to 12 languages, including English, German, Spanish, French, Hindi, Italian, Japanese, Korean, Mandarin, Dutch, Portuguese, and Russian.

The finance industry is becoming increasingly important and complicated, with its own language that must be treated specifically by speech-to-text software. To address this challenge, Autonomous Speech Recognition (ASR) engines like the one from Speechmatics have developed finance-specific English language packs to improve accuracy and reduce errors. These packs are crucial for industries such as contact centers, voice assistants, and banking, where accurate transcription is vital for customer experience, compliance, and regulatory requirements. The engine's new features, including Entity Formatting, help turn words into numbers, currencies, phone numbers, dates, and other financial data, making transcriptions more accurate and readable. With its finance-specific language pack and Entity Formatting, Speechmatics' ASR provides a viable solution for finance use cases, enabling businesses to save time and money through automation while maintaining compliance with regulations.

The company Speechmatics held its first full gathering since before the pandemic in July, celebrating 164 employees and a significant increase in staff over the past few years. The event saw the announcement of new features and updates to their Autonomous Speech Recognition platform, as well as recent investments that have helped propel the company forward. The CEO Katy Wigdahl reflected on the importance of togetherness and team unity, drawing inspiration from the England National Team's victory in a football tournament and emphasizing the value of being calm under pressure and authentic in leadership. As the summer comes to a close, Speechmatics encourages readers to prioritize time with loved ones and colleagues, believing that this will lead to happiness, inspiration, and success.

To stay at the summit of the speech-to-text mountain, a company recognizes the need to make its award-winning Autonomous Speech Recognition engine (ASR) accessible and easy to use for everyone, especially those without prior knowledge of how to use it. The company has created a helpful guide with five easy steps to get started with Speechmatics, covering deployment options such as virtual appliance, containers, SaaS, and hybrid, as well as two offering options: batch and real-time. The user can choose from various features like entity formatting, notifications, speaker diarization, and more, and select their preferred format and language, which includes 48 languages. The guide aims to be simple and easy to follow, allowing users to easily implement the Speechmatics API with a free trial and access to the best support for their business needs.

At Speechmatics, the company prioritizes accuracy above all else in its Autonomous Speech Recognition (ASR) engine, focusing on improving it year over year. The team uses self-supervised learning to combine speed and accuracy, reducing manual transcription errors and allowing for more efficient training with large amounts of labeled data. This approach enables the ASR to produce high accuracy levels, even for underrepresented voices such as African American voices, making communication more accessible and bridging cultural gaps. With a singular focus on innovation and accuracy, Speechmatics aims to remove communication barriers and foster a sense of understanding, ultimately leading to a more harmonious existence.

Self-supervised learning is an approach to machine learning that uses data itself to create labels and generate insights without relying on human supervision or external feedback. It has the potential to bring AI closer to human-like learning through observation, association, and abstraction. Our award-winning Autonomous Speech Recognition (ASR) engine uses self-supervised learning to improve its understanding of voices by training on vast quantities of data, including 1.1 million hours of audio. This approach enables our ASR to develop a more comprehensive understanding of voices and build common sense through incremental learning, ultimately making it more accessible and human-like.