Risky Speaking

Dellville, Arizona.

Back home from D.C., Casey sipped her coffee while watching her neighborhood quail on their early morning migration. Bending over her deck railing, she counted nine remaining fledglings bordered by echoing parents. These familiar noisy parades marked hourly breakpoints of shared existence around her home. Those well organized and cheery quail neighbors always reminded her to nurture younger counterparts.

Today meant listing catch-up items on an old-fashioned Steno notepad. Item #1: “Meet with young “syster” Brittany to work on her science project, “The Victorian era loom industry, Ada Lovelace, and symbolic computing”.” Item #2: “Get coffee creamer!”, as she winced at her bitter drink.

Inside the French doors off the broad deck, her iPhone Marimba ring-tone alerted a surprise call from Detective Gordon Swank, the police IT cyber guy. As she mastered computational snooping over the past 3 months, he’d become her contact with the National Terrorist Fusion centers. The local force was upgrading its IT skills. However bumbling detectives had failed to stick any further accusations onto that despicable stockbroker wife murderer.

Pressing the speaker button and dropping the Steno pad on the kitchen counter, Casey expected changes in today’s plans. His conversations always began formally.

“Ms. Hawk, we have a baffling problem that requires high tech advice. Are you available for some consulting?”

This underpaid civil servant often recounted how much he enjoyed scouting her earlier tips at lavish local country club golf events. His High Tech Detective career was beginning to depend on her regular assistance. And, she was intrigued with his different world view.

Her early morning voice quivered. “Of course, and call me Cassandra, Detective. It’s good to be in contact again. What’s up?”

Detective Swank’s teasing attitude took over. “Ditto, I’m learning a lot of tech stuff from our Geek Goddess, as we call you, ha! You know we have these Neighborhood Watch and drug tip lines, right?”

Casey stuck her cooling coffee in the microwave. “Yes, that’s how we citizens keep you busy. I do my part!”

He chuckled. “Indeed, you do. Strictly speaking, we should respect caller anonymity. But what the heck is this new tip stream? The caller uses a dark phone, one of those disposable jobs, and roams around our massive county. But what’s weird is the voices he or I or they or it, whatever, use.”

Casey thought she knew where this was going.

“For a while we were hearing a muffled tone, common when people are shakily scared to speak to us. Sometimes the caller sounded male, other times female, occasionally very young. One speaker was even vaguely ethnic, which is quite unusual here in Whitey-World. Then a staff assistant noted the resemblance to that iPhone Siri speaker, and we realized the caller was using a fake voice.”

Raising her fist to celebrate her premonition, Casey cheered, “Wow, Detective, that must have been really baffling.”

The detective gulped, “Those vocal sayings were more pleasant and natural than my annoying GPS. It scolds my wrong turn by saying ‘Recalculating, turn right in.3 miles’. I hate that!”

He paused. “Actually the voices are very understandable after we replay a few times. Dammit, there’s always something new.”

Casey needed to hurry him up so she could go to the bathroom. “Are the callers saying something important?”

The detective groaned, “Definitely! Our drug squad is intrigued. Those daily tip drops are about 50% on target when it comes to chemical creations, you know, often called ‘bath salts’. Sleazy shops and the usual dealers have made a comeback selling such disguised substances, even after we passed so many laws and busted dozens a year ago. Nobody cares about this lowlife informant’s identity, but, wow, how does that S O B gets so much good information?”

Casey was tired. “Come on, Gordon, I explained to you how Whisperers let me snoop so easily. Other folks have similar tricks.”

He said, “Yep, I forgot how you work. Anyway, The same person, probably, is even more interesting when it comes to that long term, never ending murder case. Certain tips are fingering some important community people, like city council members and charity heads and a sporty lawyer. Knowing how this person ties into the community might break open the case. We still believe the guy in jail is the murderer, though.”

Wearily, he sighed, “So many tips, so little time. This informant must be younger to fit with the drug user age group, while the murder case tipster is AARP age, like you. Oops, sorry. It’s all very confusing and new to us.”

Casey prodded, “So, what’s the problem if you’re getting good information, even if the information is spoken in an unusual way?”

The detective lowered his voice. “Actually, our commanding officers think this foretells what the intelligence agencies warned about back in 2007. The British called it ‘vocal terrorism’. Voices that could fool professionals like us might be used for false public announcements or in delicate negotiations.”

He ruffled some pages. “Here’s what really worries us. One local middle school is a hotbed of bored rich kids. Imitating the school principal in an emergency situation could fool the staff and security officers and cause public panic.” He sighed, “Or, maybe, a killer might round up students for slaughter. In the midst of chaos, a familiar sounding speaker would command action.”

His voice distanced as if he’d set his phone on the table. “Dammit, I just spilled coffee on an important case. I hate paper. So, anyway, this is like a national emergency test to build profiles for people who could or would use these synth voices, whatever you call them.”

Casey chuckled to herself. “Really, Gordon, do you think our government could be behind this? Could it be the Russians? OMG!”

The Detective resumed his unofficial attitude. “Yes, maybe good old Homeland Security is testing us cops before awarding a grant or just to see how we react. With our department reputation for mishandling physical murder evidence, maybe we can do better with data. Who knows how to do this voice thing? how hard is it? does it cost a lot?”

Casey was intrigued. “Are you also interested in these callers’ demographics? Like is this a kid gamer thing? Do you really need to identify your informant? Or are you just getting experience with a new possible attack vector?”

Detective Swank put her on hold then resumed, breathlessly. “Cassandra, you’re brilliant. Yes, we want to be prepared to defend our fair city. Please spend a few hours gathering examples to tutor me to set up my own speaking imitator. If this weren’t like terrorist thinking it would be fun and games. How about I come over in two days to take the voices out for a spin? Gotta go, we have another fast food restaurant robbery in progress!”

Later that morning, travel letdown hit Casey hard. The OMG Reunion wasn’t just a bust, it shook her belief in the contributions of her computing profession. Her weekend’s companionship with Sally had explored the good and bad sides of web site design affecting people with vision loss. She feared that a rampage of Pink Page attacks could accelerate the downfall of the WWW. Having worked on the early ArpaNet in the 1970s, enjoyed Usenet in the 1980s, researched hypertext argumentation as the Internet commercialization metastasized into surveillance capitalism, she couldn’t imagine life without Twitter for news and myriad website’s for information, no matter how rickety their design. Tweeting off those advertising distraction, mis-informants, and privacy intruders was a game she played in her hours of personal isolation.

The CumuLinker Whisperer offered her new social channels and a means to replenish her retirement funds. Detective Swank’s call nudged her toward private investigator gigs. Her specialty could be computational snooping using her power over Whisperer programming. Society was stumbling into sinister use cases for the Whisperer’s synthetic speech.

She’d been seeking a technology challenge throughout her retirement years. Could her professional niche using Consequential Reasoning supplement her own intuition and predictive tendencies? She wasn’t just any “Cassandra namesake” archetype, she was “Casey, the oracle” who knew how to select valid and reliable test cases to probe software qualities.

Names and nicknames distracted her travel-tired brain. She’d forgotten until the reunion about her former college nickname, “Cassy”, which she’d changed to “Casey” in her mid-career return to graduate school. Modern search engines didn’t track her identity from “CHawk@ISI’ in the 1970s, through “CsHawk” on Usenet, bypassing the intermittent “ClassyCassy”, into her modern “moniker “Casey”. Only a dip into public records could reveal her complete identity and occasional financial and legal misadventures. Colleague Sally Rhodes had encountered the British saying, and comic strip, “Sally Forth”. They’d compared nicknames with identity over their wine and appetizers in Washington.

Then they challenged each other to name the “gang of X” which they’d imagined would take on those Pink Page marauders. The two differently experienced computing elders agreed to disagree whether Pink Page attacks were sexist, let alone healthy for the Web in the long run. Both were fans of TimBL, the Web’s creator, who now railed at his ugly, nasty progeny.

Both wanted a joint project. Casey would pursue attackers identities and explore their motives. Sally would develop defenses against the attackers which could also improve the quality of the victim website’s. Sally had already grabbed the domain name for their envisioned gallery of before-during-after screen shots with their critiques.They were ElderOrder.org.

She opened her Steno pad and started a mind map to sort out her questions, doubts, and opportunities.

And then, Casey’s Internet connection dropped. She fixed lunch, waiting for at least one connection to revive, suspecting a two-hour Cyber Squirrel electrical repair job. She yelled at her iPhone, “Come on, Internet, I have work to do!”

Internet Down!

DellVille AZ May 2018.

“OMG!”, Casey thought, “Has one of my predictions come to pass?”.

Her mind ran through the questions she’d dumped on Detective Swank and their last weekly coffee meeting. She’d wondered how well prepared the city was for a full-fledge communication outage. So far, the national electrical grid seemed able to repel foreign and domestic attackers. But, what were the local vulnerabilities? The Detective made a note to scout other comparable city plans and check out the county alert system, dubbed “Code Red”. That early 2018 Hawaii false missile attack announcement was a warning to all state and regional authorities.

Now Detective Swank could be dealing with impending vocal terrorism. Many citizens and in-store entertainment systems relied on local radio, which could easily turn a hijacked microphone over to another speaker. Casey predicted a simulated or real attack within a month, perhaps at the annual rodeo.

She had her own immediate problems. Of course, the Internet goes out just when you have bills to pay and messages to return, not to mention catching up on the latest civil unrest over Congressional elections.

Casey reviewed her prior experience. Was this outage due to her flaky home router again? A call to the cable company information site would tell her if the network was down in her part of town. She couldn’t believe what was happening at that moment. “What’s this busy signal? Ok, I’ll call a friend. Amazing, there’s more buzz-buzz-buzz. The cell towers must be congested while land and cable phone systems are down.”

A quick trip to the bank ATM was high priority after her little vacation. She drove into chaos in her bank’s parking lots. The ATM station was cordoned off. Then, she was horrified to find that the building was under Lock Down! Other patrons were banging on the bank’s doors and windows. “WTF is going on?”, shouted a woman taking a selfie of her credit card before the “Temporarily Closed” sign.

Returning to her car, Casey saw her occasional taxi driver, Red, picking up an elderly lady, a town gadfly. Rapping on his window, she could tell he was agitated. In his gruff New Jersey accent, he told her what he knew.

“Phones are down all around town, both cell and land lines. Most commercial payment systems are wacky. Nobody knows what’s going on. Even calling the non-emergency police line gets buzz-buzz. Nobody has received that emergency alert call promised months ago.”

Rolling up his window and backing up, Red saluted Casey. “We taxi drivers are rallying to pick up stranded passengers. You can bet on Red to the Rescue! Gotta go.”

Now she had a trustworthy report from an observer of the town’s minute-to-minute activities. With too little cash and a non-operational debit card, her coffee creamer and refrigerator food refresh would have to wait.

Back home, her cell phone still made no calls and had no Internet connections, ditto for the cable Internet and phone. Going on 3 hours incommunicado was getting nerve-wracking. Could this big outage be related to the “vocal terrorism” attack she’d just discussed with Detective Swank, who might have more information? First she tried a text out of town to let her brother know she was OK, but incommunicado.

On a whim, she dropped by to see detective Swank at the local police station, hand carrying a thumb drive as an excuse for interrupting his work. In their cluttered office, one staffer was monitoring the emergency channels in other states. The youthful-looking police chief was muttering that outages had been occurring sporadic across the country. She showed them a graph of normal daily down-times for a few minutes versus a pattern of several hours in medium-sized rural cities across several states. The Chief didn’t want to start any rumors, but she’d alerted the tribal community members they were likely embroiled in a real attack, not another test. Detective Swank waved Casey out the door so official business could take place. She knew he’d update her at their next Tuesday morning coffee get-together where she could share expert reports from the venerable Computing Risks Digest.

Casey returned home freaked out from the news about a wave of outages. After two more hours, with her humble radio recharged, — she received reports about a vandalized trunk line, now repaired. But the repair work destroyed criminal evidence at the damage site deep in the desert.

She reviewed the situation. “Our attacker might be a lone wolf or group observing our pattern of disrupted services. Certainly, this couldn’t be a single point of infrastructure failure? Was this an original, or a copycat, crime? There could be a case study project to interest National Academe infrastructure researchers.”

Slowly, the Internet came back on for both smart phone and laptop. Her Twitter timeline showed no national warnings of attacks elsewhere. The long-awaited local message finally announced “Emergency services are experiencing intermittent interruptions. Keep trying if you need 911!”.

What a day! She had helped define a new terrorism threat defense in the morning and experienced a possible terrorist attack in the afternoon. As the shadow of a towering monsoon thunder storm passed through, weather alerts rang out on her phone and radio. She realized the usual Internet and Web information sources weren’t worthy of trust. Did this Cassandra-aka-Cassy-aka-Casey really have predictive powers?

Now that her search engines were back in action, she could start her “vocal terrorism”assignment for Detective Swank. Her recent weekend in “Sally World” would pay off. That well-adjusted, visually impaired colleague had showed her the VoiceDream app on her iPhone. They played with voices that sounded bi-coastal or like newscasters weaned from fly-over Country. Simulated ethnic vocal characteristics didn’t fit the vocabularies of articles they were asked to speak.

Older, raspier voices were non-existent. Sally resurrected “ATT Claire” on her decade-old desktop. However, the first major synthetic voice producer, ATT, was no longer competitive with NeoSpeech, Nuance, Accapella, and Microsoft Mobile. TalkBack and VoiceOver had energized the text-to-speech market. Foreign accents were available incase a terrorist wanted to stir up immigration issues. Casey was amazed that the Apple Voice Store was so rich in files that each cost less than an iced mocha with whipped cream.

A little research on forensic speech recognition suggested that it would be easy to distinguish human-spoken from synthetic-read recorded passages. Humans breathed!Casey noted that the point wasn’t whether a calm person or an algorithm could identify a synthetic terrorist command but rather how inattentive crowds react to commands heard from a disembodied entity. The commercial success of synthetic voices had de-sensitized listeners to tone and focused them on important content that might not be repeated.

But what about those country and generational accents? Would an attacker need to clone a specific voice?

“No problem,” Sally texted back, “algorithms do that! people losing their voices to illness record themselves for researchers to capture their speech traits. There’s a T.E.D talk and an Economist podcast on vocal cloning. Maybe it’s yet another an AI claim, but I personally would trust the Scottish and US labs that define the speech slices to be catenated into smooth wordings.”

Sally pointed out another possible clue to distinguish synthetic from human voices. Print disabled people using text-to-speech over the past four decades had reported a type of error known as a “Speak-o”. Multi-syllable words like “Pakistanis” were mis-pronounced as “Pa-kis’-tan-is”. Synthetic speech generators didn’t know parts of speech or tense for words like “read”. Sally claimed that her years of print-disabled listening enabled her to tell real from synthetic voices, but she doubted most people had that skill. Alas, her wonderful assistive technology could turn into yet another weapon for gun-fueled disturbed citizens.

To complete this easy assignment, Casey recorded some threatening tweets, transcripts from past terrorists, offenders from the “Me Too” lineup of 2017, and nuggets from the press conferences of The Great Trickster and his spokesperson du jour.

She finished her report by noting that Pink Page attackers could easily replace a link to a recorded message with an MP3 of synthetic speech sounding like the declared speaker. Everybody could be listening to “Take news” in yet another trick of technology and another gift of The Great Trickster to society.

Casey jotted down a list of the top five most likely targets for vocal terrorism. The events weren’t far off in time or place.