Nervous System: Rethinking Voice Recognition at the Canada Border

Before the 1990s, residents living near the US–Canadian border had a problem. Agents staffed the border only for limited business hours, and people frequently would find themselves trapped in the wrong country overnight when cross-border engagements lasted longer than expected. This was a long-standing inconvenience to the local populations—until a clever use of biometric technology saved the day.

The border problem was especially acute along western states such as Montana. For example, citizens of Scobey, Montana, had to contend with a locked, impenetrable barrier that descended across their community at 6:00 p.m., or 9:00 p.m. in the summer months, to separate them from the town of Coronach, Saskatchewan. Anyone attending a concert or movie, a funeral or wedding, or a sporting event, or even going on a shopping trip, could become stranded away from home. Even bad traffic could have punishing consequences. The nearest 24-hour border involved a 120-mile detour, and although some citizens resorted to driving that distance to make it to their own bed, this was not viable for farmers whose crops extended across the border. Instead, farmers had to forgo the ability to tend cross-border crops until the convenience of the border guards’ shifts permitted them to cross.

Because of these acute pressures in this remote town of barely more than a thousand people, the Scobey–Coronach border was chosen as the site of a daring experiment.

US President Bill Clinton and Canadian Prime Minister Jean Chretien signed an agreement to implement the first-ever fully automated border gate. Starting in 1996, residents could use a biometric authentication system to facilitate after-hours crossings. Biometrics are ways of mathematically representing certain physical characteristics of a person in electronic form and can be used as a means of verifying the person’s identity.

To use the new “Automated Permit Port” system, residents needed first to register with the border to get a special pass card. Then, the person would select a secret pass phrase and enroll their voice print in the automated system. Last, the person would be issued a four-digit PIN code.

According to the 1998 Congressional publication Alternative Technologies for Implementation of Section 110 of the Illegal Immigration Reform and Immigrant Responsibility Act of 1996 at Land Borders, the Automated Permit Port operated as follows:

An enrolled applicant who seeks admission at Scobey after normal port hours enters a pre-assigned Personal Identification Number (PIN) into a kiosk at the port. This prompts the system to call up the individual’s record. The applicant then recites a pre-determined phrase into a telephone receiver at the kiosk and the system compares the stored record and the phrase being recited to confirm identity. If a match is made the applicant is admitted without the intervention of an officer and a record is made of the entry.

The first person to pass through the automated gate was farmer and movie theater operator Edgar Richardson, who lived in Scobey but maintained farmland in Canada. Richardson was not the first of many, however. The Scobey-Coronach border is one of the least used ports of entry between the US and Canada. Data from the US Bureau of Transportation Statistics indicate that in 2015, fewer than twenty vehicles crossed this border daily.

Although the Scobey–Coronach Automated Permit Port was a pioneering implementation of biometrics to facilitate an entirely unmanned border gate, it was not the first use of voice recognition technology for border crossings. In 1995, the US Immigration and Naturalization Service had begun testing a combination of voice recognition and facial recognition software systems to verify the identities of pre-enrolled persons crossing the US–Mexican border at California’s Otay Mesa (San Diego). The Otay Mesa test system entailed registered users to install a transponder in their vehicles that could transmit images and sounds of the drivers and occupants to a computerized verification system as cars drove through a dedicated lane. The goal was to establish a reliable method of identification that would permit pre-enrolled users to pass through without having to stop, while freeing border officials to focus more attention on interrupting drug smuggling and other criminal activity.

Even accounting for how new voice recognition technology was in the late 1990s, the initial reports indicated that the systems were performing well. The Canada Border Services Agency (CBSA) reported that the voice recognition technology it used had 95 percent accuracy in recognizing the speaker from a previously enrolled voiceprint. The software also was said to have the ability to compare unrecognized voices to existing enrolled voiceprints to identify if the unknown speaker was a Canadian or an immigrant.

Despite the advantages to residents and the apparent success of the technology itself, all border security practices were reevaluated in the aftermath of the September 11, 2001, terrorist attacks. The automated voice recognition systems were decommissioned, and residents were once again obliged to cross at the convenience of in-person agents.

This article was originally published in Legaltech News on November 2, 2021. The views and opinions expressed in this article are those of the author and do not necessarily reflect the opinions, position, or policy of Berkeley Research Group, LLC or its other employees and affiliates.