“I need the help of the United States Government to answer the requests of the group that has held me for three and a half years. And please help me at home. Thirty three years of services to the United States deserve something. Please help me.”
Former FBI Robert Levinson — Proof of life (c. Sept. 2010)
“In the fields of observation, chance favours only the prepared mind.”
Louis Pasteur — University of Lille, France (December 7, 1854)
April 02 2020 — On December 9 2011, the Levinson’s family released a hostage video they had received in November 2010. A forensic analysis of the Audio could reveal clues about his location at the time. Do you feel like playing real-life Sherlock Holmes today? You could be the one person who notices a sound no one else has recognized previously. So, let us get started! Follow us on Twitter: @INTEL_TODAY
On December 9 2011, Levinson’s family released a hostage video they had received in November 2010.
In that video, Robert appears to have lost considerable weight, and repeatedly pleads for help in returning home.
The video is — unfortunately — rather short, and does not appear to give a clue about the location except for some music that may have been intended to deceive the FBI.
Captured ex-FBI agent Robert Levinson pleads for help
Getting to Know Your Tools
Feel free to use your ears but an old scientist may prefer to use mathematics. Specifically, a Fourier transform of the audio file will reveal the frequency distribution of the signal.
I could not find a “spectrum analyser” online but it is rather straightforward to code a short program as there is plenty of information (Python, MATLAB) available.
You want to be able to perform the following tasks:
Extract the Audio file in a suitable format
Read, Write, and Play the audio file
Select specific parts of the Audio file
Fast Fourier Transform [FFT] and visualize the data
All of these functions exist, and you only have to put them together.
Before doing any analysis, you will make sure that the code is working properly. So, I recorded a few notes — from A3 to A6: 220, 440, 880 & 1760 Hz — and FFT the audio files.
Our tool is working just fine and we are ready to analyse Levinson’s audio file!
What to look for?
Each one of us is unique. Each one of us has a unique set of skills derived from life, school, habits, hobbies, workplace…
So my advice to you is to think of a sound that others may not be aware off.
For instance, you may not know that mains electricity (also known as household electricity, street power, and grid power.) is a source of noise.
And, of course, that sound frequency is related to the utility frequency.
In large parts of the world this frequency is 50 Hz. In the Americas and parts of Asia it is 60 Hz. Notice that I said “related”, not equal to the utility frequency.
In practice, the signal will show up at twice that frequency, i.e. 100 and 120 Hz, as well as at higher harmonics.
So, if we find no sound at all at these frequencies, Levinson may have been kept in a cave in Afghanistan.
Now, a 50 Hz signal would be compatible with a location in Iran or Pakistan, but a 60 Hz signal would point to Saudi Arabia in that part of the world.
Forensic Analysis of Levinson’s Audio File
So, after reading the Audio file, we can look at the whole signal. Remember that mp3 has a sampling rate of 44.1 kHz and the Levinson’s audio file is about 59 seconds long. So we have about 2.6 10^6 samples.
Notice that near the end, there is a rather long pause as Levinson becomes emotional and is no longer able to speak.
So let us analyse the date during that second of silence (from 2.06 10^6 to 2.12 10^6).
A signal at 100 Hz is clearly visible. So we may conclude that Levinson was kept in a location where house power is available and the frequency of the grid is 50 Hz.
“Chance favours only the prepared mind”
There is simply no way to predict what could be a useful clue and who could be the right person to discover it. Then again, chance always favours the prepared minds.
Let me give you an extreme example. Do you believe that one could ever deduce the ambient temperature from an audio?
Entomologists study insects. Some of them research the Biology of Insect Song. The songs of certain species can actually be used to approximate the temperature.
A well-known example is the Snowy Tree Cricket. Here is a short video in case you are not familiar with the song of this fellow.
Counting the number of chirps given in 13 seconds and then adding 40 gives a close approximation to the actual temperature in degrees Fahrenheit.
Likewise, similar formulas can be determined for a variety of species of singing insects.
The Fugitive — The “L Train” Scene
Disappearance of Robert Levinson — Wikipedia
Seeking Robert Levinson — FBI – FBI.gov
Basic Spectral Analysis — MathWorks
audioread — MathWorks
Voice frequency — Wikipedia
Mains hum — Wikipedia
Electrical network frequency analysis — Wikipedia
Measurement of the mains frequency — European Grid Online Monitor
Robert Levinson — Forensic Analysis of Audio File Could Reveal Clues