The Use of Acoustical Forensic Evidence in High Profile Criminal Cases

The Use of Acoustical Forensic Evidence in High Profile Criminal Cases

Edward J. Primeau, CCI, CFC


This article uses four major United States historical events to explore and apply forensic science and its progress regarding acoustical forensic evidence. The learning objectives in the following article are to help the reader better understand the role of Acoustical Evidence for gunshot assessment as well as the overall concept of acoustical forensic evidence.The author has chosen four major historical events of the 20th century that have been the objects of conspiracy theorists and provided the historical foundations of each event, followed by a discussion of the use of acoustical forensic evidence for each event. Serving as case studies for the following paper are three assassinations, John F. Kennedy, Robert F. Kennedy, and Martin Luther King, plus the 18-minute gap in the Watergate tapes.

Acoustical Forensic Evidence is an audio observation that is made using original acoustical information or recorded acousticalinformation that is the result of an event like a gunshot, accident, catastrophic event, or a variety of audio recordings that are to be used in a variety of civil litigations, like a concealed recording. Recorded acoustical evidence may be provided by law enforcement and require forensic analysis and evaluation in order to determine the series of events that occurred and are not otherwise non-documentable.

This article presents a summary of each historical event, followed by opinions expressed by the author of the current state of Acoustical Evidence for gunshot assessment.  The following sections will present the author’s opinions about how these forensic principles played a role in the investigation of each one of the four crimes under review.

Learning objetives:

This article gives readers an insight into forensic acoustical evidence for gunshot assessment and how that evidence is interpreted through scientific theories that can be documented. It is possible for forensic experts to have different interpretations when examining the same acoustical evidence. This article is particularly appropriate for criminal justice personnel and litigators, as it introduces the power and potential of assessing, investigating, and interpreting acous¬tical evidence and its relevance to criminal activity.

Another learning objective is to help readers understand that acoustical sounds are sus-ceptible to precise scientific measurement, that they are explicable and modelable,and that their number, timing, echo, and acoustic signature can have significant forensic value.

Readers also will learn how acoustical forensics contributed to the investigation of four major United States crimes of the 20th century.

Published in The Forensic Examiner - August, 2015


The second half of the 20th century was a time of great political upheaval in the U.S. From 1963 to 1972, America witnessed three assassinations and a presidential cover-up. All of these events were initially considered the work of single individuals acting alone. However, years of investigation have spawned numerous conspiracy theories in an attempt to fully account for the testimony and forensic evidence in each case.

In this paper, the author will attempt to separate the acoustical evidence in each case: the assassinations of John F. Kennedy, Martin Luther King, and Robert F. Kennedy, as well as the 18.5 minute gap in the Watergate tapes. In preparation for examination of this acoustical evidence, I will present a summary of the current state of Acoustical Evidence for gunshot assessment.

I. Fundamentals of Non-Speech Acoustical Analysis

In their seminal paper presented in November 2011 at a meeting of the Acoustical Society of America, the authors (Beck, Nakasone, and Marr) succinctly sum up “The Forensic Gunshot Problem” with three statements:

  1. Gunshot sounds are made up of one or more discrete acoustic events.
  2. The waveform characteristics of any event depend on many different variables [introduced at one of three stages]: Source, Channel, and Receiver.
  3. Most gunshots recorded in forensic conditions do not match standard theoretical models.

Let’s take a closer look at each of these three statements, beginning with #1.

Graphic 1: Discrete Gunshot Sounds Recordable at a 90-Degree Angle

A: Sound of firearm’s mechanism
B: Some weapons produce a bang due to gases escaping from the side
C: Crack produced by the supersonic shock wave (rifle and shotgun only)
D: Bang produced by rapidly expanding gases escaping through muzzle
E: Sound of bullet hitting the target

While there are five discrete sounds in Graphic 1, typical evidentiary recordings capture only two of them, C and D. This is largely due to non-optimal microphone placement during the inadvertent recording of gunshot sounds (rarely is the recording of gunshot sounds planned for, except under laboratory conditions).Rather, gunshot sounds are captured accidentally during casual or broadcast recording.

Derived from Claude Shannon’s original Source-Message-Channel-Receiver model, a gunshot, with no encoded message, follows a 3-stage model, as in Graphic 2.

Graphic 2

Graphic 2

Note that noise and distortion can be added to the waveform at every one of the three stages of transmission/recording. This will make analysis of the original sound more diffi­cult. As in the communication of a message, there may be destructive and/or non-destructive noise.Accounting for the many limitations and variables in this model is the chief chal­lenge to the forensic audio investigator analyzing gunshots.

Graphic 3

Graphic 3

Finally, we have statement #3, illustrating the difference between forensic results achiev­able under ideal conditions versus those achieved under real-life conditions. The forensic acoustic investigator must explain the recorded evidence as if it were recorded under ideal conditions and then present (according to the theoretical model) those factors that account for the deviations from that model, as found in the evidentiary recording.

For example, Graphics 4 & 5 illustrate how profound an effect that microphone placement can have on the recorded waveform of the gunshot. Contrary to popular wisdom, which holds that gunshot sounds are largely omnidirectional, they have been found to be highly directional.

Graphic 4 (Courtesy of BAE Systems & the FBI)

Graphic 4

Graphic 5 (Courtesy of BAE Systems & the FBI)

Graphic 5

One of the world’s most prolific experts on the topic of gunshot forensics, Robert Maher of Montana State University, emphasizes the role of microphone placement when he says, “Finally, it is important for audio forensic examiners to recognize that the differences in level and waveform details between on-axis and off-axis recordings of the same firearm are often significantly greater than the difference between two firearm types at the same azimuth.”

II. Assassinations and Conspiracies

Graphic 6 illustrates the chronology of the four events discussed in this paper. It is useful to note that in all four incidents, initial investigations pointed to a single perpetrator acting alone. Yet, with the passage of time, responsible investigators have become increasingly confident that all these crimes were committed not solely by the individuals charged, but by one or more other/additional persons. Furthermore, in each case, significant acoustical evidence came to light years after the event, further weakening initial conclusions.

Graphic 6

Name Date of Event Location New Evidence

John F. Kennedy


Dallas, Texas


Martin Luther King


Memphis, Tennessee


Robert F. Kennedy


Los Angeles, California


Watergate Recording


Washington, D.C.


The discussion that follows is not meant to advance any particular conspiracy theory or investigator. Instead, the goal is give an overview of the following phenomena common to high-profile investigations: initial conclusions are frequently wrong; evidence is with­held on the basis of national security; new evidence continues to surface years after the crime; acoustical evidence, as always, is imperfect and subject to interpretation; advances in acoustical forensics provide additional insight into established evidence; and new and sophisticated modeling techniques help provide a fuller scientific explanation of why and how the evidentiary recording deviates from the theoretical model.

III. John F. Kennedy

On Friday, November 22, 1963, at 12:30 P.M., President John F. Kennedy was assassinated in Dealey Plaza, Dallas, Texas. The presidential motorcade turned off of Houston Street onto Elm Street, and as it passed the Texas School Book Depository, shots rang out. The president was fatally shot by a sniper as he rode in his motorcade with his wife, Texas Governor John Connally, and the Governor’s wife.

A commission established by Lyndon B. Johnson, known as the Warren Commission, launched an investigation on President Kennedy’s death that lasted ten months.  It was concluded that President Kennedy was assassinated by Lee Harvey Oswald.

It is possible that no other crime in history has generated as many spoken and written words. Not count­ing articles and research papers, there are currently more than 8,000 full-length books on the assassination of JFK. Despite the seeming surfeit of information relating to the assassination, almost every claimed fact has spawned one or more counterfactuals. To this day, there is no generally accepted final result to the ongoing investigation, and the JFK assassination remains the great granddaddy of assassination conspiracies. The three major official findings are as follows: Warren Commission Report (9/24/64) - Lee Harvey Oswald was the sole assassin; House Select Committee on Assassinations Report (7/29/79) -the assassination was probably carried out by two shooters, including one on the grassy knoll; and National Academy of Science Panel Report (5/14/82) - there is no reliable evidence of a second gunman on the grassy knoll.

Insofar as the acoustical evidence contributes to these conclusions, we have a classic case of well-founded opinions between top-level experts. How is such discord possible? As stated in Graphic 3, actual evidentiary recordings rarely match the theoretical models used to interpret such recorded sounds. The more ways the evidence can be interpreted, the more likely a variety of forensic conclusions will result from various forensic experts.

Perhaps as controversial as the official findings, which contradict each other, is the fact that while the assassination occurred in 1963, there was no acoustical evidence uncovered until 1977. Recovered from the archives of the Dallas Police Department, the break­through acoustical evidence was in the form of a Dictabelt recording from a Dictaphone machine, which was set up to record all radio traffic. Recording two separate radio chan­nels in use by the DPD, the machine was VOX operated (voice activated).

Due to a stuck microphone PTT switch, the assassination was inadvertently recorded through the radio of one of the motorcycle policemen in the motorcade. The Dictabelt recording was turned over to Dr. James Barger of the acoustical analysis firm Bolt, Beranek andNewman, a company that had done classified analysis previously for the military. Also involved in the consultation were Professor Mark Weiss and expert Ernest Aschkenasy.

With the benefit of the photographic, eyewitness, and ear witness evidence that had been documented in the years between 1963 and 1977, the group planned to use the Dictabelt recording to tie all the evidence together and produce a convincing explanation of key details of the assassination. The group developed an ingenious plan to recreate the acoustic events in Dealey Plaza on that fateful day.

On Sunday August 20, 1978, Dr. Barger and his team closed off the Plaza and placed 36 different microphones every 18 feet along the motorcade path. Six shots were fired at a mock limousine as it traveled, and two shooting locations were used: the Texas School Book Depository (Oswald’s presumed position) and the grassy knoll to the front and right of the limo. The total of 12 shots, each recorded by 36 microphones, pro­duced 432 acoustic samples. These samples, containing unique echo and muzzle blast signatures, were to be matched with the recorded gunshot sounds on the Dictabelt.

On September 11, 1978, Dr. Barger testified before the House Select Committee on Assassinations and presented his findings.  There were four shots fired at the president, the first two from the Texas School Book Depository, a third from the grassy knoll, and a fourth from the Texas School Book Depository. Barger calculated the certainty of each shot, and he testified these were 88%, 88%, 50%, and 75% respectively. (Low-fidelity microphones, radios, and recorders precluded extensive FFT waveform matching and adaptive filtering. Furthermore, AGC circuits in the motorcycle radios further distorted the impulse signals.)

Dissatisfied with the 50% result for the grassy knoll shot, the House Select Committee on Assassinations asked Weiss and Aschkenasy to check the data independently. Finding a total of 26 echo-producing objects in the acoustic field, they recalculated with greater accuracy and ultimately found a 95% likelihood that the third shot came from the grassy knoll. In subsequent consultations, Dr. Barger agreed with this result.

Since that time, a 1982 study by the National Academy of Sciences denied the possibility of a shot from the grassy knoll. In the past 30 years since that study, various meta-studies have looked at the Barger data and reached a wide variety of conclusions. There remains a strong suspicion among many researchers that the CIA and other government agencies have a vested interest in maintaining the findings of the Warren Commission report, and as a result, the full truth is unlikely to ever be revealed. In 2012, House Assassinations Committee Chief Counsel Robert Blakey said, “I think our conclusion was correct. On balance, I say there were two shooters in the Plaza, and not just because of the acoustics. I find on balance that the earwitness and eyewitness testimony is credible.”

IV. Martin Luther King

Though no acoustical evidence exists in the King assassination, there are striking simi­larities between the King and Kennedy assassinations.In both cases: leaders who were against the Vietnam War and supporters of the civil rights movement were assassinated in a southern U.S. city; a seemingly incompetent loner was immediately charged with the crime (James Earl Ray and Lee Harvey Oswald); normal protocols, providing extra pro­tection for the visiting dignitaries, were changed at the last minute; substantial evidence proves that both were targets of CIA and FBI investigations and “dirty tricks”; and eye­witness and earwitness accounts put shots, flashes, and smoke in locations other than the presumed sites of each shooter.

In Memphis, forensics experts were never able to match the slug to the recovered rifle, nor could they demonstrate the trajectory of the bullet to be from the bath­room of the rooming house or from the brush outside (several witnesses contradict the findings of the House Select Committee on Assassinations,which stated the shot came from the bathroom).

Both the 1963 and 1968 assassinations predate the advent of portable video camcorders. With a cell phone/video recorder carried throughout the day by most Americans nowa­days, it is unlikely that future investigations of assassinations and bombings will suffer from such a dearth of recorded acoustical evidence. Though Ray died in prison, the King family continues to believe in the innocence of James Earl Ray.

V. Robert F. Kennedy

The assassination of Robert F. Kennedy took place at 12:15 A.M. on June 5, 1968, in Los Angeles, CA. The senator had just completed his victory speech following the California presidential primary. He was leaving the Ambassador Hotel through the kitchen when he was shot three times by the alleged assassin, Sirhan Sirhan, a young Palestinian/Jordanian immigrant.

Because the event had ended, the cameras (film and video) were not rolling during the assassination. It was not until 2004 that a tape recording of the assassination itself, made by an independent reporter, was discovered. Stanislaw Pruszynski was a Polish journalist freelancing for the Montreal Gazette, and he was equipped with an audio tape recorder and hand-held microphone the night of the murder. He was unaware that his recorder was still running as he followed Kennedy to the hotel’s kitchen.

Following three years of study, audio expert Philip Van Praag in Tucson announced three highly controversial findings after analyzing the Pruszynski recording ( First, computer analysis revealed a total of 13 gunshot sounds. Second, two of the shots were followed so closely by another shot that it would be impossible for the .22 revolver to fire fast enough (so called “double shots”). And third, five of the 13 shots had acoustic signatures (resonance peaks at 1,600 Hz) un­like the other eight. Van Praag posited that these five shots had been fired from a second gun, a Harrington & Richardson .22 caliber (matching that of the security guard next to Kennedy) and not Sirhan’s weapon, a .22 Iver-Johnson Cadet.

Experts have various opinions on all three findings. However, when it comes to the number of gunshots (eight, ten, or 13), Van Praag has said, “The primary point that people need to recall is that the capacity of Sirhan’s gun was eight shots. He did not have time to reload. And so anything more than eight shots prove there had to be a second gun firing in that kitchen pantry.” Van Praag adds that the eyewitness testimony of 12 people proves that the fatal bullet, fired from only one inch behind Kennedy’s ear, could not have come from Sirhan’s gun.

Van Praag presented his findings and new technique (Frequency Selective Integrate Loudness Envelope Evaluation) before the American Academy of Forensic Sciences in 2008.His report stated, “Initial study of the recording involved the conventional use of time and frequency domain waveform analysis. These methods were useful to determine timing locations of suspected muzzle blast shot sounds. These locations were correlated with what was known from the crime scene through substantial video and audio coverage before and after the shooting, together with eyewitness testimony. The use of spectro­graphic imaging was helpful in differentiating the less prominent muzzle blast sounds, but – partially due to the high ambient noise levels – was not useful to differentiate between weapons. By extracting multiple high frequency slices over the entire gun­shot interval, with each slice representing a quite narrow frequency band, and then overlaying these individual slices, subsequent slice comparisons revealed markedly differing amplitude levels of some shot sounds. This measured phenomenon was con­sistent with the positioning of two known guns within the kitchen pantry of the Ambas­sador Hotel in Los Angeles, the scene of the crime. The resulting identification of which gun fired which shots was also consistent with the known number of shots fired from the Sirhan weapon, and the timing sequence of Sirhan’s shots as recalled by the man who first apprehended him.”

VI. Watergate Tapes

On June 20, 1972, President Nixon discussed the Watergate break-in for the first time with his chief of staff, H.R. Haldeman. Unbeknownst to all but a few White House staff­ers, all oval office meetings were secretly recorded. According to the President’s secre­tary, Rose Mary Woods, she accidentally erased about five minutes of that part of the tape during the transcription process on September 29, 1973.

In November of 1973, Nixon ordered an investigation into the 18.5 minute gap found on the tape. By July 1974, Nixon had lost his battle to keep thousands of hours of recordings private. Revealed on a tape from June 23, 1972, was Nixon’s order to prevent the FBI from further investigating the Watergate break-in on the grounds of national security. Once made public, this admission forced Nixon to resign his presidency on August 9, 1974.

It was subsequently determined that the 18.5 minute gap had been caused by 5 - 9 sep­arate erasures made on a Uher 5000 reel-to-reel audio tape recorder, not the original recorder, a Sony TC-800B. The Uher was one of Nixon’s personal machines, and so the question soon became, “What was in those 18.5 minutes that was so incriminating that the President personally erased them?”

Administration held an open competition for anyone who could retrieve the lost conver­sation from the erased tape. Ultimately, five entrants were given access to the original archived tape.

Paul Ginsberg of Professional Audio Laboratories in Spring Valley, NY, was among those chosen. Ginsberg owned a working TC-800B and created an in-depth profile of the ma­chine’s electronic and mechanical characteristics. In this way, he was able to more easily differentiate between taped sounds and machine artifacts. Recorded on 1/2-inch audio tape at 15/16-inches per/second, the erasure contains pops and buzzes but no recognizable speech.

Ginsberg loaded the audio into his PCAP II Plus forensic audio processing workstation. Used by the FBI to clean up the audio of surveillance tapes, the software included high-, low-, notch- and comb-filters, plus numerous equalizers (comb-filters are useful in reduc­ing harmonic noise, such as the 60-cycle hum on the Nixon tape; and inverse comb-filters enhance harmonic components, especially vowel sounds, rendering voices more intelligi­ble). After months of work, Ginsberg was unable to identify or retrieve any vocal sounds from the erased part of the tape.

James Reames of JBR Technology in Woodbridge, VA, was another entrant, who also owned a classic Sony TC-800B. He followed many of the protocols used by Ginsberg but also looked for incriminating fingerprints on the tape using a custom mirror assembly on the recorder.

Reames also examined the tape using an array of magneto-resistive read/write heads, such as those used in computer hard drives. Designed to float just above the magnetic record media, these heads can read the smallest changes in the depth and direction of magnetic particles. Unfortunately, he was unable to detect any vocal sounds, either.


Beck, Steven D., Nakasone, Hirotaka & Marr, Kenneth W (2011) An introduction to forensic gunshot acoustics, presented at the Acoustical Society of America, San Diego, CA.

Committee on Ballistic Acoustics, National Research Council (1978) Reexamination of acoustic evidence in the Kennedy assassination, Boulder, CO.

Koenig, Bruce E. (1983) Acoustic gunshot analysis, The Kennedy assassination and beyond, in FBI Law Enforcement Bulletin, Washington, D.C.

Maher, Robert C. (2010) Overview of audio forensics, in Multimedia analysis for security applications; H.T. Sencar, et al, eds. Springer-Verlag, Berlin, Germany.

Maher, Robert C. & Shaw, Steven (2008) Deciphering gunshot recordings presented at AES 33rd International Conference, NYC, NY.

McNichol, Tom (2007) Richard Nixon’s last secret. Wired, San Francisco, CA.

Manchester, Phil (2010) An introduction to forensic audio, in Sound on Sound, Cam-bridge, U.K.

O’Sullivan, Shane (2008) RFK Must Die: Pruszynski Recording Reveals Second Gun documentary video (, on Documentary Channel, Nashville, TN.

Pepper, William (2003) An act of state, The execution of Martin Luther King. Verso Pub-lishing, NYC, NY.

Scally, Chris (2013) Unsound acoustics?, in JFK, Echoes from elm street; M. Bridger and B.Keane, eds. Cambridge Academic, Cambridge, UK.

Summers, Anthony (1998) Not in your lifetime. Open Road Integrated Media, Inc., NYC, NY.

Wikipedia (2014) United States House Select Committee on Assassinations (http://


About the Author

Ed PrimeauMr. Primeau, founder and CEO of Primeau Forensics, Limited and author of That’s Not My Voice! has been practicing audio and video forensics for 30 years. He is a member of AES, ACFEI, and IAI and is a Certified Criminal Investigator (CCI) and a Certified Forensic Consultant (CFC), as well as a member of the American Board of Recorded Evidence with the American College of Forensic Examiners International. Visit for more information.

Publisher Dr. Robert O'Block, American College of Forensic Examiners.

© 2015 Primeau Forensics, LTD