1 State-of-the-art CVRs are beginning to employ frequency shift key (FSK) encoded data in CVR audio in accordance with internationally developed standards。 However, many in-service CVRs lack this FSK component。

Figure 1。 Time scale alignment issue; black shading indicates events common to the CVR and  FDR。

0 for not keyed)。 Microphone keying is most often sampled every one second。 This record of keying in the CVR and FDR forms the basis of timeline  alignment。

The CVR microphone keying identification accuracy and the FDR microphone keying sample rate pose an optimization problem for the synchronization of the timelines。 The problem can be stated as an optimization of two or more series of chronological events where the events are sampled with different accuracies or rates。

The problem can be further generalized by considering additional realities of recording devices and the forensic environment。 One reality is the accuracy of the device recording and playback rate (‘‘timebase’’)。 That is, a 60- minute period may be recorded in 58 minutes or 62 minutes, or more generally a tolerance of ¡ r, where r is the tolerance of recording timebase。 Further, r could be a variable, influenced by power fluctuations or environmen- tal acceleration factors that may be experienced in an accident。 While r was a significant factor in tape based, legacy recorders, today’s solid-state recorders have r9s with small variations over time。 For example, US regulations require a tolerance of ¡ 0。125% per hour, (Federal Aviation Administration, 2010)。 This equates to ¡ 。75 seconds for a 10 minute recording or ¡ 6。75 seconds for a 90 minute recording。

Another reality of the forensic environment is that CVR and FDR microphone keyings may be mixed with other events that have greater accuracy or forensic    importance。

For example, a critical warning may be recorded by the CVR and FDR and be added to the event timeline。 Further, there may be missing events between the two timelines, for example, a short microphone keying could be heard on the CVR that was not sampled by the FDR (i。e。, for microphone keyings with a total duration of less than one second)。

Literature Review

Combining the CVR and FDR timescales is an essential step in the investigation process。 The procedure is most often accomplished by an investigator analyzing the audio of the CVR for the start and stop of radio transmissions and recording this numerical information。 The series of transmissions are then compared to a series of microphone keyings recorded on the FDR。 These events can then be correlated to other time series events, such as air traffic control。 (Brazy, 2009; National Transportation Safety Board, 2010; O9Callaghan, 2009; Smith, 2009)。 Text summarizing the analysis is often presented as  follows:

The times reported in the attached CVR transcript are Eastern Standard Time (EST), and represent the time that each comment or sound begins。 Time is specified to the nearest whole second, unless otherwise  noted。

The CVR and FDR data were synchronized to one another by comparing the FDR ‘‘Key VHF’’ parameter with radio transmissions as heard on the CVR recording。 By comparing the CVR elapsed time (time since the beginning of the CVR recording) for radio transmissions to the corresponding FDR Subframe Reference Number (SRN) for ‘‘Key VHF,’’ a relationship between the CVR elapsed time and the FDR SRN time can be developed。 Generally, a single keying event can be used to synchronize the CVR to the FDR to within ¡ 1 second, due to the FDR9s 1 Hz sample rate for ‘‘Key VHF。’’ Using multiple keying events may increase the accuracy of the synchronization。 In this case, the start and end times for six radio  transmissions  (12  keying  events)  were  evaluated。