Financial institutions turn to speech recognition technologies to combat mis-selling, prevent fraud and offer a pathway to enhanced customer service

By Alex Fleming, Products Manager at Speechmatics

Stick to the script!

The financial services sector is one of the most highly regulated in the world. With the advent of General Data Protection Regulation (GDPR) and the massive fines associated with the violation of data protection, regulatory compliance has never been more important. In 2018, The European Union’s GDPR introduced fines for groups of companies of up to €20m or 4% of annual worldwide turnover, whichever is greater – far exceeding the previous fines. The tsunami of PR around the protection of data has also meant that these issues have come into the public eye. End users are more and more affected and influenced by the negative press companies receive when their capability to protect their customer data is called into question. For this reason the punishment for any kind of compliance oversights or mistakes far surpass just the financial implications impacting brand reputation and trust. Organisations are looking automatic speech recognition to enable better understanding of the data. In the case or pre-recorded calls the capability to audit and discover elements within that data which could pose a potential compliance risk and to do some automatically reducing human intervention and at scale.

ASR is not limited to pre-recorded content. High performance real-time ASR enables dynamic monitoring to ensure that their representatives deliver information and advice in line with evolving compliance regulation no matter their industry or region.

Interaction history

Calabrio claim that most contact centres only evaluate 1-2% of the calls in the contact centre. Quality assurance teams are under increasing pressure to ensure that caller data is protected as well as the issues raised are handled with immense quality and efficiency. With ASR, there is no ambiguity regarding who said what and when. Transcription means that interactions can be far more easily indexed and discovered than an equivalent audio format. This text-based format also enables the enrichment of downstream natural language processing (NLP) solutions to derive insight through analytics.

Automated call capture

Real-time transcription enables financial organisations to take compliance out of the hands of humans and move them to the responsibility of automated systems. The consequences of capturing personally identifiable information (PII) and personally identifiable financial information (PIFI) are significant. Sophisticated number recognition empowers the roll out of synthetic agents, voice bots and voice-enabled UX tools able to engage with callers and process confidential information.

Real-time

The benefits of transcribing pre-recorded content are undeniable for financial institutions. By its transformation into text, voice then becomes easily accessible, key words can be found, and insight gained. Also, areas that require improvement can be identified and any PII data that may have slipped through the net in the call recording screening can be discovered and redacted. Real-time transcription unlocks the value of voice to fuel tools to derive sentiment and emotion and other aspects held within the voice dynamically. This means agents can be fed insight and data in real-time taken from voice and help guide and steer the trajectory of the conversation and ensure elements like script adherence are completed. The ability to utilise these elements in voice mean that compliance can be monitored and even trigger additional call flows so potential risks can be both identified and averted.

Adaption for recognition of bespoke vocabularies

ASR solutions that also have custom dictionary features able to empower their users to quickly add unique terms and vocabularies unique to them and their sector mean that language models can be adapted to deliver better and more accurate results. For trading floor environments where language is very different from regular conversations, this feature not only enables accurate capture of interactions but also enables evaluation of conversations. Being able to do that can highlight potential miss-selling scenarios where code words might be used that signify a specific action or information.

On-premise

High-performance solutions that can be deployed on-premise offer enhanced levels of security and compliance for the financial sector. With data security paramount, this means that audio and any associated transcripts remain within their closed environment and abstracted from the public internet and 3rd party solutions. On premise options from ASR providers enable organisations to operate these solutions within their own environments and as part of their existing technology solution stack.

Conclusion

ASR technology has grown up significantly in that time and matured to enable it to add potential value to the financial sector. Now, the levels of accuracy provided by these solutions and the ability to deliver in real-time mean that there has never been a better time for financial organisation to incorporate speech services like ASR into their solutions. We expect that machine learning and AI-based tools will have an increased presence in financial markets. The tools offer capability to optimise and automate auditing of pre-recorded content and also monitor compliance. So really, can the financial industry afford not to deploy ASR?