Detecting lies has never been so easy
By Lim Tri Santosa
JAKARTA (JP): Do you sometimes feel like you're being lied to? For example, wouldn't you love to know if that politician on TV is really telling the truth about his proposed tax cuts? There are countless situations we face daily where knowing just who is telling the truth, and who isn't, would come in handy.
But how do you find out? There are many different forms of lie detection, most of which are obtrusive, expensive, and hard to use, and even if people would agree to be hooked up to one, they would most surely be insulted! Now, there is a much better, and much less invasive way to find out if you're being lied to.
The Voice Stress Analyzer (VSA) makes it possible to test subjects for truthfulness even if they aren't present! The VSA processes the raw input of a voice. Vocalization can be voiced in person, recorded and played back or taken off the radio or television. Yes, it's possible to analyze voice stress without the subject being present or even knowing about it.
Voice analysis
It seems that a person's voice changes under stress. In some occasions a person's voice will even "crack". These stress-caused changes are what make voice stress tests possible. The principle and application of VSA is simple technology. Researchers found that human voice frequencies in the range 8 to 12 Hz are sensitive to honesty.
Human ears can hear an audible spectrum from between 20 Hz to 20,000 Hz. Thus an 8 Hz to 12 Hz frequency is a sub-sonic spectrum, which cannot be heard by human ears. When a person is being honest, the average sound of that condition is generally below or equal to 10 Hz and usually above 10 Hz in dishonest situations.
Vocal chords vibrate in the 8 Hz to 12 Hz range. This is considered as a feedback loop, similar to the way a thermostat heater works. An average temperature is maintained by raising the temperature a little above the setting, switching off, and not coming back on until the temperature is a little below the set temperature. Just as the temperature swings up and down over time, so do the vocal chords and larynx as they seek to maintain a constant tension.
But keep in mind that the above theoretical statement is not always correct. Why? Lie detection is not an exact science, because not all lies are the same.
Technology provides the interrogator with information, but no light flashes red or green to identify whether the subject's statement is true or false.
Let me give you an example. Have you seen Bill Clinton's grand jury testimony? I think you have already heard the ending of this story, haven't you? It was very difficult to detect when the president was being truthful and when he was being "legally accurate". Nobody knows, except the president.
Once, I was angry with my 5-year-old niece because I suspected she had stolen cake from the kitchen. I asked her whether she was doing it, and she said calmly and confidently: "No, I didn't take the cake, I didn't steal it."
Finally I asked her little brother and he confessed that his sister ordered him to steal the cake, otherwise she said she would kiss him. In this situation my niece's words were "legally accurate", because in fact she did not steal the cake. Her little brother was telling the truth, the whole truth, nothing but the truth. Thus, being "legally accurate" does not mean one is telling the truth, only the correct one.
Everybody knows that stress, also known as anxiety or being uptight, can be brought on by a simple thought or memories; of a loved one's passing on, for example, or suddenly thinking of problems in the future. Some people have high average stress levels, and some have low stress levels, and averages change from day to day, along with moods.
What all people have in common is that their stress levels are constantly changing within their current range; changes which indicate the "perceived jeopardy" or danger of statements made. A lie is often dangerous, humiliating, or injurious to be caught saying, thus lies tend to stand out in the area of stress measurement.
Entertainment use
Prevaricator version 1.03 is a combination of sound wave player and recorder, which performs real-time spectral frequency analysis on audio data playback. It has standard media player controls: play, pause, stop and record buttons (almost like Windows' Sound Recorder). It is freeware and can be downloaded from http://www.softseek.com/Education_and_Science/Science_and_Engineering/D_2261 2_index.html
The user should have a sound card in their computer in order to transfer the audio for conversion into a wav file. It is amusing to analyze the interviews after recording them; users can hook up their cassette player cable into a sound-card line input, then push the play button of the cassette player and press the record button on the software.
Prevaricator will convert it into a 16-bit wav file that you can save later on. As each word is played back, the real-time graph is updated to reflect the stress patterns present in the words.
This little program resolves the maximum, minimum and average spectral frequency of the voice. The red lines represent tolerance limits in percentage from the maximum and minimum frequency. The green line in the middle represents the nominal average frequency of the audio. The red lines will give us a visual indication of frequency deviation from the normal one. Users can also check the highest, lowest and average levels of the voice's frequency.
There is a great tool to analyze the voice's frequencies. It is called a Frequency Analyzer. Every sound can be represented as a combination of sine waves of various frequencies. Mathematically, this splitting into component frequencies is called a Fourier Transform. Since the sound usually changes with time, the Fourier Transform is calculated many times per second.
Technically, the Frequency Analyzer performs a Fast Fourier Transform of the sound (users need a sound card and a microphone to use this program). How does it work? It analyzes the voice and splits it into its component frequencies. This program can graph the pitch in real time, patterns of speech, the harmonics of vowels and the noise of sibilants.
The size of the program is only 64k! It is freeware, so you can download it from http://www.relisoft.com/index.htm. Even though monologue artist Butet Kartaredjasa can mimic former president Soeharto's voice, the frequencies are different from the real ones. Each human voice has unique frequencies, even though the human ears perceive it as the same voice.
Obviously, VSA can be a valuable tool to assist in analyzing opinions of public speakers or politicians, and the things they said. We should consider the context in order to determine the truth. It could be anger, fear, pride, or guilt.
Lie detection is actually a matter of measuring certain physiological characteristics of the body and interpreting whether they are caused by the stress of lying. But not all stress is caused by lying. The above software is just a simple tool for entertainment use only. The real analyzer is our conscience mind.