Voice channels - the risk you’re missing?

Join Daniel Yates, Global Relay Voice Surveillance SME, as he uncovers the reasons why capturing voice channels is now a compliance necessity – not an option.

Voice should now be a focus for firms

Voice has long been a blind spot in communications compliance. Regulators have placed no specific enforcement pressure on firms around voice compliance, leaving them with little incentive to focus on it.  However, as trust in compliance technology has developed, so have regulatory expectations, leading to voice coverage becoming an urgent priority. While manual review of thousands of hours of calls has been a historic challenge, new AI-enabled surveillance technology allows firms to capture, transcribe, translate, and analyze voice channels to identify real risk.

Technology can offer a good solution for surveillance purposes

Voice capture technology has evolved

Large Language Models (LLMs) have been transformational in allowing firms to capture voice communications thanks to their highly accurate transcription and translation capabilities. Earlier generations of speech-to-text were rigid, struggled with accents, financial terminology, and multi-language use. LLM-enabled transcription now means firms can accurately generate transcriptions of real-world conversations and enable communications to be analyzed with their vital context.

Accessibility to voice data is so much more mature now

Streamlined setup, seamless integration

Historically, deploying a voice compliance solution was a significant infrastructure project, with long implementation timelines, heavy IT department involvement, and bespoke integrations. Now, with a more mature ecosystem, whether a firm is using Teams, Zoom, Webex or Cisco, the friction of getting connected has dropped considerably. Ongoing solutions management is more straightforward and intuitive, allowing compliance team to make changes directly.

A lot of these systems can now be integrated very easily from the get-go

Better data, better outcomes

With capture solutions ensuring complete, high-quality data from every call, conversation, and channel, teams are able to search and reconstruct conversations quickly, leverage lexicons to flag risk, and reveal patterns human reviewers would struggle to identify at volume, Firms are moving beyond reactive compliance, using voice data to understand potential financial and non-financial risk across their business, and are able to act accordingly.

A lot of these systems can now be integrated very easily from the get-go

The biggest misconceptions when it comes to voice channel compliance

Many see voice as inherently harder to capture than electronic communications channels, but this gap has narrowed significantly. This is in part thanks to advances in capture and transcription technologies that enable voice monitoring and risk detection to exist in the same compliance-tech ecosystem as text-based communications.

The second misconception is that firms may assume their existing eComms archiving solution is giving them coverage, even though it is not. Voice is its own data type with its own capture, retention, and retrieval requirements. Treating it as an afterthought, or assuming it’s handled by your existing compliance tech stack, can leave firms exposed to risk.

A lot of firms believe that capturing written communications is enough, but in actual fact it is not?

Which risks can voice monitoring identify?

Voice conversations can include warning signs of both financial misconduct, like insider trading, and non-financial misconduct like bullying or harassment – a growing area of regulatory focus. Advances in AI-enabled transcription and translation are enabling firms to navigate challenges with jargon or “trade speak,” issues with audio quality, and language barriers to effectively identify risk.

While not an “active” risk, regulatory expectations that firms capture and archive all required communications data are high, and firms that overlook voice data may find themselves under regulatory scrutiny.

People believe that using voice communications channels doesn’t leave a paper trails, but this is really not the case anymore

The future of voice compliance

Firms are already moving away from keyword-based flagging and lexicon-based models to LLM-enabled models, where the system can learn what “good” and “bad” looks like in context and can flag risk based on sentiment, rather than relying on a static list of trigger words.

Next is near-real-time intervention capability, which moves compliance from reactive to preventative. Voice communications happen in real time, which means risks manifest in real time – and compliance should be able to keep pace. As AI becomes more embedded in how firms communicate, the compliance function will need to evolve to capture and oversee those interactions too.

We have some big advancements in technology coming, in terms of being to identify users in the call which will make the whole surveillance workflow so much easier and streamlined