New technology perfected by Stephen St.Croix is recovering long-lost sounds. It might even bring back Watergate’s famous 18 1/2-minute gap.
New technology perfected by Stephen St.Croix is recovering long-lost sounds. It might even bring back Watergate’s famous 18 1/2-minute gap.
The National Archives in Washington recently received an extraordinary request from Stephen St.Croix, who is the president of a company called Intelligent Devices. St.Croix is an expert in the esoteric field of sound extraction, and he proposes to solve one of Washington’s most enduring mysteries: what was on the 18 1/2-minute gap on the Watergate tape? He believes he can retrieve the sound from a tape that may have been erased as many as five times more than 25 years ago.
The very notion of St.Croix’s vocation conjures images of a world of paranoid but talented ex-spooks twitching with suspicions. To find out just how possible it was to dig, archaeologically, in the realm of old sound, I called Intelligent Devices to talk to St.Croix. Days passed before he returned the call, and when he finally did, my 5-year-old daughter picked up the phone (right after I did) and then clumsily hung it up.
“Is someone listening in on this conversation?” St.Croix demanded. When I explained what had happened, he paused for a moment, as if to determine whether my explanation was just a fast-thinking coverup. Then he seemed to relax.
“You know, if you’re paranoid and they are listening,” he said almost cheerfully, “then you’re not paranoid.” St.Croix agreed to let me visit him, but because of security concerns, I was told to come to his house, not the office. He didn’t want me to publish the name of the town he lives in except to say that it’s outside Washington. He gave me precise instructions for the cabby. I was told to get out at the end of a certain cul-de-sac.
“Then wait for the cab to leave,” he said. “I’m serious. And after you’re sure the cab’s gone, walk down the driveway to the left. Don’t come to the front door. Just keep walking. You’ll set off the lasers in my woods. I’ll know you’re coming and come out to meet you.”
It went off without a hitch. As I approached a side door, a trim, well-built 52-year-old St.Croix, dressed in black boots, black pants and a black shirt, stepped out to greet me. He has the eyes of a young Gerald Ford and close-cropped gray hair, which is a new look for him. His driver’s license from just a few years ago shows a biker guy with about two or three feet of blond hair. St.Croix escorted me to his living room. He rocked in a black leather chair and talked while playing an unamplified electric guitar.
St.Croix was a session player for most of his life, and his studio wall holds a platinum record for Stevie Wonder’s album “Songs in the Key of Life.” In the mid-1980’s, he gravitated to the lucrative field of audio cleanup, restoring and improving the sound for movie classics like “The Wizard of Oz” and “Gone With the Wind.” St.Croix may be a handsome biker-surfer dude with lots of modern toys, but in his heart he’s a sound guy, a techie. It took a while for him to channel his inner dweeb. But once he appeared, St.Croix was quite happily talking the shoptalk of a sound geek who thinks extracting hidden voices from fuzzy tape is just really cool.
He loaded his G3 Mac with Speech Extraction System, the software his company sells, and fed a segment of audiotape into the computer. I snapped on my headphones and heard nothing but roars and crackles with some kind of human voice deeply buried in the cacophony. “These are the tools one works with,” he explained, pointing down the side of the work screen. The menu bar offered Noise Reduction and Bandwidth Limiter and Harmonic Notch and Tone Removal and Adaptive Extraction. He explained that the recording was an authentic piece of tape made by a reporter aboard Air Force One a few years ago.
“This feature cuts out the rumble of the airplane,” he said. By clicking the button and moving the cursor, you can remove all the sound that is naturally outside the register of the human voice. By just lopping off mostly bass sound and a bit of treble, I could almost make out the words being uttered in what was obviously a conversation between two men. St.Croix ran the remaining sound through the Adaptive Extraction module, which “listens to the audio and rewrites its own code to better fit what it hears.” As he cycled the sound through repeatedly, the sound got better and better. Pretty soon, I could hear President Clinton and President Bush talking to reporters as they flew to the funeral of Yitzhak Rabin.
That was a pretty easy problem to solve. To show me something trickier, St.Croix loaded a recording from a case he had worked on a while back. “It’s a drug dealer arranging a pickup,” he said. “He’s talking on the telephone, but he installed sound distorters that completely mask the conversation.” I heard a variety of irritating mechanical noises, although threaded deeply in the background was the inaudibly basslike “mmmm”‘s of human intonation. It was as if the speaker were talking underwater with his mouth closed.
St.Croix ran the segment through the Tone Removal system. Tones are sounds that maintain the same frequency and volume. Human language is highly intonated; that is, it is constantly changing frequency and volume. The Tone Removal module analyzes the sound, recognizes the frequencies that are constant and then strips them out. It cleaned up the sound enough for me to tell there was human conversation on the tape, but it was still unintelligible. Then St.Croix began to loop it through the Noise Reduction module. It never reached broadcast quality, but what began as a dense murmur became an audible voice saying: “Take the package and go back and drop it off. . . . Meanwhile keep an eye out for other people. Don’t talk to anybody. If anybody asks, don’t panic. You don’t know anything.”
Until recently, recovery efforts were limited by analog sound. For a technician to cut out a piece of bad sound, it had to be “notched” out the way you chop at a tree. While you could get the bad sound out, you also had to remove parts of the sound you wanted to keep. With analog, there was a built-in limit to any recovery job. But computers changed all that.
“Digital allows you to make square cuts,” St.Croix said. He demonstrated how an unwanted piece of sound on one part of the sound curve was right beside some speck of voice. Then he digitally extracted the unwanted noise, like slicing down a length of fabric with a razor blade to remove a single thread. The computer allowed him to do this repeatedly — hundreds and thousands of times, removing microscopic bits and flecks of tones and rumbles and hisses. The crackle and hum slowly faded as the coherent conversation of human voices rose to the surface.
It was early morning and St. Croix wanted breakfast. He took about five minutes to load a fanny pack with some electronic gear. St.Croix is a gizmo fiend. He has a feeling of mechanistic intimacy that probably dates back to a long-ago reading of “Zen and the Art of Motorcycle Maintenance.” His house is packed with dust-free objects of high tech, and in his driveway he has three race cars and half a dozen motorcycles. We climbed into his cherry red Corvette — fully customized by St.Croix with a radar detector that can even warn of cops in helicopters — and rumbled up the driveway. St.Croix flipped open his cell phone like Captain Kirk and, addressing someone or something, explained that we were leaving and to “arm” the house. He turned out of the cul-de-sac and punched it.
At a local diner, St.Croix explained his Watergate theory. Audiotape, he said, is basically rust painted onto a piece of tape. “Iron oxide has many interesting properties, it turns out,” he said. “It catches and releases oxygen very easily. That’s why we have iron in our blood, to regulate oxygen flow. We’re basically rust, too. But on a piece of tape, if you could look at it closely, you would see an array of randomly placed iron particles. What a tape recorder does is set these in a particular order that can then be reproduced as sound.”
Erasing a tape simply takes the patterns set in the iron-oxide particles and scatters them randomly so that, when played, what you hear is the familiar hiss of blank tape. In the past, several sound technicians have had a go at the Watergate gap, but what they tried to do was retrieve the sound by listening at the outer edges of the tape. The theory was that the eraser head was a bit smaller than the recording head, so that some sound might have been preserved at the hem of the tape. St.Croix operates from a different theory.
“No erasure is complete,” he said. “The head rescrambles the sound filings but usually only gets about 95 percent. Even if they erased it numerous times, each time they only really get about 95 percent. Which means that there is a possibility that buried in there is the original sound.” St.Croix proposes to build a special tape machine with 200 separate heads that would read the piece of quarter-inch tape. The computer would be programmed to eliminate the randomly situated particles and read only the ones that, on each vertical strip passing through the heads, are the same; that is, the particles that represent the remnants of coherent sound.
To persuade officials at the National Archives, St.Croix is proceeding with an experiment — he’ll record on an old tape from the 70’s and erase its content several times, then try to recover that sound. If it works, a committee at the archives will then vote on whether to give St.Croix a shot at the Watergate tape and the opportunity to excavate the secret that Nixon took to his grave. We know from written logs that it is a conversation between Nixon and Bob Haldeman on June 20, 1972, three days after the break-in. Nixon’s accusers have always assumed that the gap was really a recording of the two men planning the coverup. Nixon’s defenders have always supported the president’s claim that this swath of tape was accidentally erased when Nixon’s secretary mistakenly hit a pedal-triggered erase button with her foot. Sound technicians are dubious of Nixon’s claim about his secretary because within the gap can be heard the clicks of someone turning the machine on and off about five times. That’s a lot of mistakes, they argue, unless the secretary’s foot was temporarily seized by palsy.
Sound ends up as heat,” St.Croix mused later in the morning. “We speak, and the sound waves vibrate the molecules in the air, bounce off the walls and vibrate the molecules some more but ultimately dissipate as friction, as heat.” And increasingly, he said, those waves are getting captured somewhere.
“When you walk down the street, there are any number of video and even audio devices pointed at you,” St.Croix said. “In an urban environment, it wouldn’t be hard these days to reconstruct a conversation you thought you were having privately walking down the street.”
One of the more peculiar theories about sound was developed hypothetically some 30 years ago. A technician named Richard Woodbridge III coined the phrase “acoustic archaeology” in the August 1969 issue of Proceedings of the I.E.E.E., the engineering journal. Woodbridge theorized that there were many occasions when sound might innocently get scooped out of the air and preserved. For example, when an ancient potter typically held a flat stick against a rotating pot, he was accidentally (and crudely) recording into the clay the sounds around him. Woodbridge wrote about experiments he performed pulling basic noises off a pot. Another experiment involved setting up a canvas and then talking while making different brush strokes.
“This is to record the finding of a spoken word in an oil portrait,” Woodbridge wrote. “The word was ‘blue’ and was located in a blue paint stroke — as if the artist was talking to himself or to the subject. Parenthetically, the search was long and tedious. The principle, however, was established.”
St.Croix listened to my recounting of Woodbridge’s theories on our way home from breakfast, and fishtailing at 80 miles per hour, he said to me, “Let’s don’t go overboard.” He then cocked his cell phone. “Disarm,” he whispered, and without encountering any cross-fire, we pulled peaceably into his driveway. Safely nested back within the sanctuary of his electromagnetic web, St.Croix began to recall some rather Woodbridgean work he had done on “The Wizard of Oz.” He explained that there were certain audible peculiarities if you listened carefully to the master sound reel. For instance, you might hear a piece of music that involved a flute and a timpani, yet most of the timpani bass sound was too low in frequency to be recorded by the equipment available in 1939, when the movie was made. But the recording of the flute had a mild Doppler effect, as if something were making the flute throb in and out. St.Croix realized that this phantom effect was a result of the lost bass parts of the timpani.
“So it was a kind of sound scarring,” he said. “We analyzed it and were able to go back and restore some sound at the limits of the audio — some bass, some treble — that wasn’t even there in the master recording but was there in the original recording session.”
As we walked up his driveway through the leaves (and lasers), St.Croix called me a cab on his cordless phone, and then we sat down and waited for it, chatting idly. Considering the piece of plastic in his hand, he said: “This is just a standard cordless. But I have had it especially encrypted. The phone is good and safe.” He looked at the receiver admiringly. “It’s kinda nice.” We stretched out our legs on the asphalt. The early afternoon light was just beginning to angle. There was nothing left to say. Sunbeams shot majestically through trees whose orange-brown leaves drifted down continuously in a familiar October rustle. We listened to the sounds of autumn. They have never sounded so loud.
Author: Jack Hitt
News Service: New York Times