Speech-to-text has come of age

RMA Journal, The, Sept, 2002 by Frank DiLorenzo

"Efficiency Quotient" is a series designed to help institutions maximize their efficiency of operations and revenue per employee. Topics range from credit analysis, to loan production, to technology.

Imagine...

Imagine going on customer or prospect calls and, in the car on the way to your next meeting, dictating what just transpired into your digital recorder. When you get back to the office, you plug the digital recorder into your PC and download your meeting notes into your speech-to-text software, which, in turn, automatically types your call memos, using the word processing software you've specified. While this is happening, you return phone calls, respond to e-mails, call new prospects, review financial statements, and so forth. Your call reports are done before you know it. All you have to do is proofread them and e-mail them to the appropriate parties and departments (record retention, credit analysis, the senior lender) and you're done. The best part is that you don't have to imagine it. This technology is real, it's affordable, and it works.

Speech-to-Text Technology

When speech-to-text technology became commercially available in the late 1980s, the software was costly, slow, and inaccurate. With today's processing speeds and memory capabilities, however speech-to-text software packages are available for under $200, work with normal, continuous speech, have vocabularies of more than 300,000 words, and have a degree of accuracy that generally meets or beats 98%.

Speech-to-text technology in commercial banking can enable a bank to achieve several goals:

* Keep personnel who need to know customer/ prospect information well informed.

* Provide better information to bank examiners.

* Assist with the preparation of credit analyses.

* Give account officers more time to perform more new-business-development calls.

These benefits should translate into the bank booking additional business and result in a meaningful improvement in earnings.

Desired Features

When choosing a speech-to-text software package that's right for you, look for the following features:

* Easy setup and tutorial software.

* A high-quality headset microphone.

* Compatibility with whatever business software suite your institution uses.

* Mobile recording capability with a compatible portable digital recorder.

* A vocabulary of 300,000-plus words, with an emphasis on business and finance.

* A typing speed of up to 160 words per minute.

* Dictation shortcuts to insert commonly used text.

* The ability to recognize voice commands for tasks such as bolding, underlining, etc.

* Voice commands that let you search the Internet, solve math problems, schedule appointments, and track contacts.

* The ability to ignore the "aahs" and "umms" you might utter while dictating.

* Speech recognition accuracy of 98% or higher.

* The ability to recognize numbers, dates, and dollar amounts and have them typed as such. For example: "One thousand dollars" would be typed as $1,000.00.

* A toll-free help line and/or online support for the product.

Technical Requirements

Minimum technical requirements for speech-to-text software generally include the following:

* Windows 98, Windows Millennium, Windows 2000, or Windows NT 4.0.

* A processor performance equivalent to an Intel Pentium III or greater.

* 128 MB RAM.

* 250 MB free disk space.

* A sound board supporting 16-bit 22 KHz recording.

* CD-ROM drive for installation.

* Speakers (for some tutorial software).

If you bought your PC within the past 18 months, the chances are good that it already comes equipped with these minimum system requirements.

To maximize efficiency with the system, you also should purchase a compatible digital recorder. Some digital recorders can provide as many as 22 hours of recording time. The digital recorder should come with a USB cable to attach to your PC for direct downloading of its recordings to the speech-to-text software. You can also get a digital recorder with a voice activation feature, so that if you're not speaking, it's not recording. This is ideal for hands-free operation. Normally, for a few dollars more, you can also purchase a noise-filtering microphone that provides clearer recording in noisy environments (such as in a moving vehicle).

Impact on Productivity

What does speech-to-text capability mean for the Commercial Loan Department? In a typical environment of more account officers, not known for their typing skills, coping with fewer available administrative personnel, speech recognition technology can go a long way toward improving the department's efficiency. Benefits include:

* More accurate and timely preparation of call reports.

* Easier dissemination of loan request terms and information to credit analysts.

* An increase in account officers' calling time. After all, they are not typing reports.

* Smoother bank exams. Examiners' questions concerning relationships can frequently be answered by reading the call reports.

* Faster ramp-ups. In instances of account officer turnover, a new officer will be brought up to speed with a relationship much more quickly by reading the call report history.