Financial Institutions’ Demand for Quicker and More Accurate Voice Transcription Is at an All-time High, Says RegTech Provider Steeleye
- 23.05.2022 12:15 pm
- Most communications surveillance tenders now ask for transcription services, and not just for fixed-line calls but also for MS Teams, Zoom WhatsApp and more
- Increased regulatory scrutiny of communications rules are driving firms to seek faster and more accurate transcription services
- SteelEye has developed an AI-driven Financial Services Transcription Engine in response to this record demand, which is 120x faster with 90%+ accuracy
SteelEye, the compliance technology, and data analytics firm, has reported record demand for voice transcription services. The firm analyzed all Requests for Information (RFIs) for communications surveillance received over a 12-month period and found that in 100% of cases, businesses listed voice transcription as a key requirement. Firms are seeking voice-to-text solutions for calls on fixed-line, mobile, MS Teams, Zoom, WhatsApp, and more to power their communications surveillance programs.
The high demand is being driven by increased regulatory scrutiny of communications rules on the back of the Covid pandemic and the shift to flexible working – prompting firms to seek faster and more accurate transcription services. In the UK, the FCA has repeatedly reinforced the need for financial firms to record and monitor relevant telephone and electronic communications, noting that WFH can lead to an increased risk of misconduct.
We have also seen increased scrutiny of firms that fall short of oversight rules in the US, even though it is not yet a requirement to capture and store voice communications. In December, the SEC and CTFC handed out a $200m fine to JP Morgan for failures related to employee communications monitoring. Since then, the SEC has also opened an investigation into Citi Group, Goldman Sacs, and Morgan Stanley over employee communications.
Existing options fall short
The demand for better and more automated transcription of voice data can be seen as a move from the industry to get ahead of regulators by enhancing their communications surveillance programs.
Transcribing and indexing voice communications can enable firms to better analyze their data to identify risks and signs of wrongdoing. Yet most incumbent voice transcription services are slow with poor accuracy, making the speech-to-text process lengthy and the output often unreliable, impacting the analytics that can be derived:
“One of our clients explained that it used to take them over 24 hours to process the voice communications that occur within their firm each day,” said Matt Storey, Chief Product Officer at SteelEye. “Where data volumes increase, so does the processing time which is where the problem arises for large firms who end up racking up large transcription costs with limited upside. Running searches using inaccurate transcription data is ineffective and leaves firms vulnerable to missing key signs of risk.”
To address the transcription technology gap and enable faster and more accurate voice analytics within financial services, SteelEye has released an industry-specific transcription engine. The SteelEye transcription engine is 120x faster with 90%+ accuracy, powered by DeepGram, whose speech recognition model is AI-driven to yield enhanced results.
“The demand for better technology in communications surveillance is only set to increase. Financial services firms tend to follow the herd when it comes to innovation. We are already seeing tier one firms investing in advanced and real-time transcription as a priority. If regulators start to expect the digitalization of voice as a norm, the “herd” will follow,” says Brian Lynch, President of SteelEye Americas.
SteelEye’s transcription model is tailored to the financial markets with linguistic customization that accounts for industry jargon and the specific language used to manipulate markets, amassed from hundreds of court filings covering market abuse offenses. With better voice-to-text accuracy, surveillance models can become more intelligent, allowing firms to better identify compliance risks whilst reducing false positives.