2017-09-19, Dan Nyberg, Forensiker, Informationsteknik
Ljudanalys vid NFC
2
2017-09-19
Innehåll
• Kort genomgång av vad NFC gör
• Vad kan man göra med ljud?
Vad gör vi vid NFC
• Hörbarhetsförbättring
– Filtrering
– Nivå kompression – Brusreducering
4
2017-09-19
Vad gör vi vid NFC
• Äkthetsundersökning
– Auditivanalys – Visuellanalys
– Tekniskundersökning
Vad gör vi vid NFC
• Ljudjämförelse
– Jämför omstridda ljud mot olika alternativ – Likheter och skillnader etc.
6
2017-09-19
Vad gör vi vid NFC
• Förbedömning talarjämförelse
• Säkring av ljudmaterial
• Inspelningsförfaranden
• ENF-jämförelse
• Talarkonfrontation
Vad kan man analysera från en
ljudsignal?
Jämföra rumsakustik
Avståndsbedömning Skottriktning/Triangulering
Jämföra röster Förbättra hörbarheten
Leta klipp-punkter
Tidsbestämning ENF
Jämföra punktljud
Bestämma inspelnings- utrustning
8
2017-09-19
Olika typer av material
• Telefonsamtal
• Dolda inspelningar
• Referensinspelningar
• Etc.
Faktorer som påverkar
• Ljudnivå på ljudkällan
• Vind, Tempratur, Luftfuktighet, Ytor, Hinder
• Inspelningsutrustningens återgivning
• Automatisk förstärkning, (Automatic Gain Control)
• Kodning – eventuell ”utsmetning” av ljudet
11
2017-09-19
Datastrukturer
• Header
• Container
• Metadata
• Vad händer när filer sparas?
– Skillnader – Likheter
13
2017-09-19
Maher Acoustical modeling of gunshots
0 0.02 0.04 0.06 0.08 0.1
-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
Time [sec]
Amplitude [Linear Scale]
Figure 8: Modeled gunshot recording, first 100ms, with simulated reflections for geometry depicted in Figure 5,
except with barrel pointing east rather than west; .357 Magnum handgun.
The comparison of Fig. 7 and Fig. 8 shows the similarities and differences between two simulated recordings in the same space, but with the orientation of the firearm reversed. If the simulation had not included the directional characteristics of the firearm, the signals depicted in Fig. 7 and Fig. 8 would be the same because the geometrical modeling is the same: including the empirically determined directional information results in a noticeably different outcome.
In the configuration simulated in Fig. 8, the firearm is pointed in the general direction of the microphone, while the reflections from the north and west correspond to the lower level off-axis direction with respect to the firearm’s barrel. For audio forensics, this comparison could conceivably allow determination of the likely orientation of one or more firearms in a monophonic audio recording, or provide information that might disambiguate the order in which two or more firearms were discharged.
4. CONCLUSION
The study reported in this paper uses the directional characteristic of a firearm and a compound image- source method to simulate the acoustic signal recorded at an arbitrary recording location. This approach takes into account the significant differences in sound pressure waveforms and levels between on-axis and off- axis reflection geometries.
As demonstrated with the two simulation examples in Section 3, the orientation of the firearm can have a significant effect upon the received waveform. This effect is not predicted by simple convolution of a single gunshot recording with a single impulse response. Thus, this work is intended to emphasize the need for forensic examiners to be wary of common assumptions that can oversimplify the potential complexity of acoustic evidence. This caveat may be particularly important for the design of automatic gunshot classification systems.
5. ACKNOWLEDGEMENTS
The assistance of Prof. Steven R. Shaw in making the original gunshot audio recordings is gratefully acknowledged. The assistance of the Montana State University Red Bluff Research Ranch, Mr. Pete Olind, ranch manager, is also greatly appreciated.
6. REFERENCES
[1] Maher, R.C. "Audio forensic examination:
authenticity, enhancement, and interpretation,"
IEEE Signal Processing Magazine, vol. 26, no. 2, pp. 84-94, March 2009.
[2] Koenig, B.E., Hoffman, S.M., Nakasone, H., and Beck, S.D., “Signal convolution of recorded free- field gunshot sounds,” J. Audio Eng. Soc., vol.
46(7/8), pp. 634-653, July/August 1998.
[3] Freytag, J.C., and Brustad, B.M., “A survey of audio forensic gunshot investigations,” Proc. AES 26th International Conf., Audio Forensics in the Digital Age, pp. 131-134, July 2005.
[4] Maher, R.C. “Modeling and signal processing of acoustic gunshot recordings,” Proc. IEEE Signal Processing Society 12th DSP Workshop, Jackson Lake, WY, pp. 257–261, Sept. 2006.
[5] Maher, R.C. “Acoustical characterization of gunshots,” Proc. IEEE SAFE 2007: Workshop on Signal Processing Applications for Public Security and Forensics, Washington, D.C., pp. 109–113, April 2007.
Maher, 2011, Audio Engineering Society
14
2017-09-19