The financial services sector is notoriously jargon-heavy with industry terms that are either completely unique to the industry or that can be confused with commonly used phrases. Acronyms such as VAT or SEC or abbreviations e.g. Generally Accepted Accounting Principles (GAAP), and the word ‘gap’ can often confuse standard speech-to-text engines. Speechmatics can now capture the speech data as intended, turning unstructured, audio data into usable information. By improving the accuracy of transcripts, downstream tasks can be more consistent and streamlined for users.
Global experts in deep learning and speech recognition, Speechmatics has built the most accurate and inclusive speech-to-text engine available. Historically, training data had to be manually tagged, classified or ‘labelled’. This has resulted in engines trained on narrow datasets, which fail to represent the diversity of voices that use them. In contrast, Speechmatics’ speech-to-text engine is trained through exposure to hundreds of thousands of individual voices using millions of hours of unlabelled, more representative voice data. This has enabled a paradigm shift in accuracy, dramatically reducing both AI bias and errors in speech recognition. Given the broad range of demographics that exist within financial services, Speechmatics’ new offering will be key to supporting and sustaining inclusivity in the sector.
Katy Wigdahl, CEO, Speechmatics, said, “Our aim is to understand every voice regardless of race, gender or accent and I’m proud that Speechmatics has overcome significant challenges that traditional speech-to-text engines have struggled with.
However, we wanted to go even further and dive into the complexities that specific industries present. Some sectors are known for complex terms and jargon that, if added to our global models, risk making the technology less effective for other users. This led to our approach for domain-specific packs that can directly address the needs of individual sectors. Financial services was an obvious place to start but we hope our language pack will set a blueprint for every high-stakes industry where the financial, reputational and social cost of misunderstanding is high.”
Customers are already using the finance language pack to transcribe financial news and earnings calls as well as utilising the technology to aid call centre analysts and traders. The pack is the first industry-specific pack and paves the way for industries with equally complex terminology such as medicine and law.
