Page 29 - Summer 2015

recorders clipped to the vests of law enforcement officers to document their interaction with the public. As recorder costs continue to decrease, it seems inevitable that more and more investigations will include evidence obtained from these portable devices and dashboard recorders (McGinty, 2015).
The ubiquity of audio recording devices will require an emerging specialty in handling, searching, and analyzing the potentially huge number of recordings available from a public incident that leads to a forensic investigation. The investigative challenge that occurred after the 2013 Boston Marathon bombing involving dozens of surveillance record- ings and hundreds of digital photos taken by smartphone and camera-toting bystanders near the finish line provides a glimpse of what is likely to occur in future incidents involv- ing video, audio, and still photo evidence (NOVA, 2013).
Transportation Accident Investigations:
The Cockpit Voice Recorder
Because commercial aircraft accidents are so remarkably infrequent, aviation regulatory agencies can afford to spend substantial resources to investigate the cause of accidents when they do occur. Arguably the most important develop- ment in accident investigations today is the invention of the flight data recorder (FDR) and cockpit voice recorder (CVR) equipment required on all civilian commercial passenger flights, large private jets, and many military flights, the so- called “black boxes” (National Transportation Safety Board, 2015).
Originally developed using special fireproof magnetic tape, contemporary FDRs and CVRs now record in digital form using nonvolatile solid-state memory. FDR systems on con- temporary airliners can record hundreds of flight param- eters, actuator positions, and sensor readouts every second, with memory capacity for up to 25 hours. Yet, even with the plethora of flight data, the acoustical information from the CVR is often indispensable for accident investigators to piece together what happened leading up to the accident.
CVRs capture four separate monophonic channels: the pi- lot’s headset microphone, the copilot’s headset microphone, a cockpit area microphone (CAM) mounted in the cockpit’s ceiling panel, and the fourth channel that often is used to record the intercom communications between the pilots and the flight attendants. Modern CVRs record up to 120 min- utes of audio in a memory buffer loop, sequentially over- writing the oldest data with new data. Thus, in the event of an accident, the audio forensic examiner will have a record- ing containing the sounds from the 2 hours preceding the crash (National Transportation Safety Board, 2007).
Analysis of the four audio channels of CVRs primarily in- volves transcribing the spoken words of the pilot and copi- lot, and any other utterances by members of the crew. In ad- dition to the flight crew's speech, the microphones can also pick up nonspeech sounds that can be very important to accident investigators. Engine sounds, airframe vibrations, avionics audible warning alarms, and sounds of cockpit in- trusions or other commotion may all have significance to the investigation. For example, the March 2015 demise of Germanwings Flight 9525 in the French Alps has been at- tributed to deliberate action by the copilot reportedly based in large part on preliminary evaluation of the CVR evidence. What’s more, even the sound of the pilots’ respirations can give information about the flight crew’s health, state of alert- ness, and level of anxiety or agitation (Stearman et al., 1997; Byrne, 2002; McDermott, personal communication).
CVR systems are typically activated automatically whenever the aircraft is powered up, whereas the FDR systems collect flight data only from the point at which the plane becomes airborne. This means that the CVR may contain important information about flight crew checklist completion, preflight discussion, and similar audio information obtained before takeoff that isn't covered by the FDR information (McDer- mott, personal communication).
Assessing the Fidelity of Audio Forensic Findings
Audio forensic findings that will end up in court are sub- ject to the same validity considerations used for other types of scientific expertise before the information can be stated in the courtroom. In 1993, the US Supreme Court affirmed in Daubert v. Merrell Dow Pharmaceuticals, Inc., that un- der Rule 702 of the Federal Rules of Evidence (which cov- ers both civil trials and criminal prosecutions in the federal courts), “The Rules—especially Rule 702—place appropriate limits on the admissibility of purportedly scientific evidence by assigning to the trial judge the task of ensuring that an expert's testimony both rests on a reliable foundation and is relevant to the task at hand” (Daubert v. Merrell Dow, 1993). Assessing the reliability of audio forensic opinions can be a challenge. Unlike DNA comparisons that can be expressed in a formal statistical sense, a forensic acoustics question such as “Is the utterance present in the evidentiary recording the voice of Suspect A?” traditionally has not been amenable to having a strong statistical basis for the opinion. Due to growing judicial skepticism about the scientific ba- sis for almost all forensic testimony in fields ranging from fingerprints and bite marks to handwriting and carpet fiber comparisons, the 2009 report by the National Academy of
Summer 2015 | Acoustics Today | 27