Speech Analytics for Financial Institutions

Financial services companies around the world are getting into the recording business in a big way. More and more, they tend to record just about everything being said in and around their organization: phone trades with clients, conference calls, audio from video conferences, and cellphone conversations, to name a few.

The main reason for this is compliance. A series of federal, state and international regulations are forcing institutions to blanket their operations with voice recording. Frequently, specific information must be captured as part of the client order handling process. Compliance officers may also need to be informed when an order is concluded by phone, but the agent forgot to mention certain keywords or phrases.

Coping with New Requirements

The rapid growth in regulatory requirements has resulted in a tremendous increase of recording activities. Banks previously recording 1,000 turrets of financial traders might now need to record and retrieve audio from half-a-million telecommunications endpoints.

This massive increase in volume has strained traditional monitoring methods, now unable to check even one percent of interactions. To stay in compliance, financial institutions are increasingly using powerful speech analytics solutions to automatically check the content of every single phone conversation.

The Power of Phonetic Analysis

Today's technologies for speech analytics can be categorized into three groups: keyword spotting (KWS), phonetic-based analytics (PBA) and transcription-based analytics (TBA).

Both KWS and PBA process voice recordings using phonetics, the building blocks of words and sounds in conversations. These fast and flexible technologies can even flag issues in real time in the middle of a call. Without taking up a lot of bandwidth, they can quickly search recorded calls for words and phrases. Phonetic analysis does not depend on any given language and can be easily implemented and maintained compared to technologies based on complex dictionaries.

The Benefits of Transcription

Neither KWS nor PBA will automatically present analysts with the text of a conversation or a "word cloud" to search for clues. Instead, the analyst needs to know the search term in advance. The context of words will not be considered.

Transcription-based analytics (TBA) instead converts conversational speech to text, using massive computing power and large reference dictionaries. In addition to spotting known keywords, new words, phrases and meanings can be detected, too.

TBA automatically creates a text version of the phone call, allowing unlimited searches, a powerful tool when trying to determine the root cause of developing situations. The context of words can also be evaluated to avoid false positives.

A text version of calls makes it possible to process data with other systems and techniques such as data mining to automatically alert analysts about new topics for trend analysis. However, TBA uses substantial processing power so it requires a considerable investment in technology and hardware as well as a concerted effort to "train" the system and maintain the dictionaries – separate dictionaries are needed for each language to be processed.

The Versatility of Speech Analytics

Integrating any speech analytics technology with contact centers and trading desks will put financial services companies back in the driver’s seat. In addition to compliance, they will have a wealth of information at their disposal to evaluate market trends, optimize internal processes and improve customer service.

Even though financial services firms are experiencing a lot of pressure to record verbal transactions of every kind and retrieve them on demand, speech analytics solutions can turn this necessity into a strategic bonanza to transform their organization.