-
Can We Identify Unknown Audio Recording Environments in Forensic Scenarios?
Authors:
Denise Moussa,
Germans Hirsch,
Christian Riess
Abstract:
Audio recordings may provide important evidence in criminal investigations. One such case is the forensic association of the recorded audio to the recording location. For example, a voice message may be the only investigative cue to narrow down the candidate sites for a crime. Up to now, several works provide tools for closed-set recording environment classification under relatively clean recordin…
▽ More
Audio recordings may provide important evidence in criminal investigations. One such case is the forensic association of the recorded audio to the recording location. For example, a voice message may be the only investigative cue to narrow down the candidate sites for a crime. Up to now, several works provide tools for closed-set recording environment classification under relatively clean recording conditions. However, in forensic investigations, the candidate locations are case-specific. Thus, closed-set tools are not applicable without retraining on a sufficient amount of training samples for each case and respective candidate set. In addition, a forensic tool has to deal with audio material from uncontrolled sources with variable properties and quality.
In this work, we therefore attempt a major step towards practical forensic application scenarios. We propose a representation learning framework called EnvId, short for environment identification. EnvId avoids case-specific retraining. Instead, it is the first tool for robust few-shot classification of unseen environment locations. We demonstrate that EnvId can handle forensically challenging material. It provides good quality predictions even under unseen signal degradations, environment characteristics or recording position mismatches.
Our code and datasets will be made publicly available upon acceptance.
△ Less
Submitted 3 May, 2024;
originally announced May 2024.
-
Point to the Hidden: Exposing Speech Audio Splicing via Signal Pointer Nets
Authors:
Denise Moussa,
Germans Hirsch,
Sebastian Wankerl,
Christian Riess
Abstract:
Verifying the integrity of voice recording evidence for criminal investigations is an integral part of an audio forensic analyst's work. Here, one focus is on detecting deletion or insertion operations, so called audio splicing. While this is a rather easy approach to alter spoken statements, careful editing can yield quite convincing results. For difficult cases or big amounts of data, automated…
▽ More
Verifying the integrity of voice recording evidence for criminal investigations is an integral part of an audio forensic analyst's work. Here, one focus is on detecting deletion or insertion operations, so called audio splicing. While this is a rather easy approach to alter spoken statements, careful editing can yield quite convincing results. For difficult cases or big amounts of data, automated tools can support in detecting potential editing locations. To this end, several analytical and deep learning methods have been proposed by now. Still, few address unconstrained splicing scenarios as expected in practice. With SigPointer, we propose a pointer network framework for continuous input that uncovers splice locations naturally and more efficiently than existing works. Extensive experiments on forensically challenging data like strongly compressed and noisy signals quantify the benefit of the pointer mechanism with performance increases between about 6 to 10 percentage points.
△ Less
Submitted 3 May, 2024; v1 submitted 11 July, 2023;
originally announced July 2023.
-
Synthesizing Annotated Image and Video Data Using a Rendering-Based Pipeline for Improved License Plate Recognition
Authors:
Andreas Spruck,
Maximilane Gruber,
Anatol Maier,
Denise Moussa,
Jürgen Seiler,
Christian Riess,
André Kaup
Abstract:
An insufficient number of training samples is a common problem in neural network applications. While data augmentation methods require at least a minimum number of samples, we propose a novel, rendering-based pipeline for synthesizing annotated data sets. Our method does not modify existing samples but synthesizes entirely new samples. The proposed rendering-based pipeline is capable of generating…
▽ More
An insufficient number of training samples is a common problem in neural network applications. While data augmentation methods require at least a minimum number of samples, we propose a novel, rendering-based pipeline for synthesizing annotated data sets. Our method does not modify existing samples but synthesizes entirely new samples. The proposed rendering-based pipeline is capable of generating and annotating synthetic and partly-real image and video data in a fully automatic procedure. Moreover, the pipeline can aid the acquisition of real data. The proposed pipeline is based on a rendering process. This process generates synthetic data. Partly-real data bring the synthetic sequences closer to reality by incorporating real cameras during the acquisition process. The benefits of the proposed data generation pipeline, especially for machine learning scenarios with limited available training data, are demonstrated by an extensive experimental validation in the context of automatic license plate recognition. The experiments demonstrate a significant reduction of the character error rate and miss rate from 73.74% and 100% to 14.11% and 41.27% respectively, compared to an OCR algorithm trained on a real data set solely. These improvements are achieved by training the algorithm on synthesized data solely. When additionally incorporating real data, the error rates can be decreased further. Thereby, the character error rate and miss rate can be reduced to 11.90% and 39.88% respectively. All data used during the experiments as well as the proposed rendering-based pipeline for the automated data generation is made publicly available under (URL will be revealed upon publication).
△ Less
Submitted 28 September, 2022;
originally announced September 2022.
-
Forensic License Plate Recognition with Compression-Informed Transformers
Authors:
Denise Moussa,
Anatol Maier,
Andreas Spruck,
Jürgen Seiler,
Christian Riess
Abstract:
Forensic license plate recognition (FLPR) remains an open challenge in legal contexts such as criminal investigations, where unreadable license plates (LPs) need to be deciphered from highly compressed and/or low resolution footage, e.g., from surveillance cameras. In this work, we propose a side-informed Transformer architecture that embeds knowledge on the input compression level to improve reco…
▽ More
Forensic license plate recognition (FLPR) remains an open challenge in legal contexts such as criminal investigations, where unreadable license plates (LPs) need to be deciphered from highly compressed and/or low resolution footage, e.g., from surveillance cameras. In this work, we propose a side-informed Transformer architecture that embeds knowledge on the input compression level to improve recognition under strong compression. We show the effectiveness of Transformers for license plate recognition (LPR) on a low-quality real-world dataset. We also provide a synthetic dataset that includes strongly degraded, illegible LP images and analyze the impact of knowledge embedding on it. The network outperforms existing FLPR methods and standard state-of-the art image recognition models while requiring less parameters. For the severest degraded images, we can improve recognition by up to 8.9 percent points.
△ Less
Submitted 3 May, 2024; v1 submitted 29 July, 2022;
originally announced July 2022.
-
Towards Unconstrained Audio Splicing Detection and Localization with Neural Networks
Authors:
Denise Moussa,
Germans Hirsch,
Christian Riess
Abstract:
Freely available and easy-to-use audio editing tools make it straightforward to perform audio splicing. Convincing forgeries can be created by combining various speech samples from the same person. Detection of such splices is important both in the public sector when considering misinformation, and in a legal context to verify the integrity of evidence. Unfortunately, most existing detection algor…
▽ More
Freely available and easy-to-use audio editing tools make it straightforward to perform audio splicing. Convincing forgeries can be created by combining various speech samples from the same person. Detection of such splices is important both in the public sector when considering misinformation, and in a legal context to verify the integrity of evidence. Unfortunately, most existing detection algorithms for audio splicing use handcrafted features and make specific assumptions. However, criminal investigators are often faced with audio samples from unconstrained sources with unknown characteristics, which raises the need for more generally applicable methods.
With this work, we aim to take a first step towards unconstrained audio splicing detection to address this need. We simulate various attack scenarios in the form of post-processing operations that may disguise splicing. We propose a Transformer sequence-to-sequence (seq2seq) network for splicing detection and localization. Our extensive evaluation shows that the proposed method outperforms existing dedicated approaches for splicing detection [3, 10] as well as the general-purpose networks EfficientNet [28] and RegNet [25].
△ Less
Submitted 3 May, 2024; v1 submitted 29 July, 2022;
originally announced July 2022.
-
Gamma-ray emission in proton-induced nuclear reactions on natC and Mylar targets over the incident energy range of Ep = 30-200 MeV. Astrophysical implications
Authors:
Y. Rahma,
S. Ouichaoui,
J. Kiener,
E. A. Lawrie,
J. J. Lawrie,
V. Tatischeff,
A. Belhout,
D. Moussa,
W. Yahia-Cherif,
H. Benhabiles-Mezhoud,
T. D. Bucher,
T. R. S. Dinoko,
A. Chafa,
J. L. Conradie,
S. Damache,
M. Debabi,
I. Deloncle,
J. L. Easton,
M. Fouka,
C. Hamadache,
F. Hammache,
P. Jones,
B. V. Kheswa,
N. A. Khumalo,
T. Lamula
, et al. (15 additional authors not shown)
Abstract:
We have measured the gamma-ray line production cross sections in proton-induced nuclear reactions on various target nuclei abundant in astrophysical sites over the incident energy range of Ep = 30 - 200 MeV. We carried out experimental campaigns in joint collaboration at the K = 200 cyclotron of iThemba LABS using a high-energy resolution, high-efficiency detection array composed of 8 Compton-supp…
▽ More
We have measured the gamma-ray line production cross sections in proton-induced nuclear reactions on various target nuclei abundant in astrophysical sites over the incident energy range of Ep = 30 - 200 MeV. We carried out experimental campaigns in joint collaboration at the K = 200 cyclotron of iThemba LABS using a high-energy resolution, high-efficiency detection array composed of 8 Compton-suppressed clover detectors comprising 32 HP-Ge crystals for recording the gamma-ray spectra. In the current paper, we focus on de-excitation lines produced in proton irradiations of natC and Mylar targets. In particular, on the prominent 4.439 and 6.129 MeV lines of $^{12}$C and $^{16}$O which are among the strongest lines emitted in solar flares and in interactions of low-energy cosmic rays with the gas and dust of the inner galaxy. We report new gamma-ray production experimental cross section data for ten nuclear lines that we compare to previous low-energy data sets from the literature, to the predictions of the TALYS code of modern nuclear reactions and to a semi-empirical compilation. In first approach, performing calculations with default input parameters of TALYS we observed substantial deviations between the predicted cross sections and experimental data. Then, using modified optical model potential and nuclear level deformation parameters as input data we generated theoretical excitation functions for the above two main lines fully consistent with experimental data. In contrast, the experimental data sets for the other eight analyzed lines from the two proton-irradiated targets exhibit significant deviations with the predicted cross section values. We also report line-shape experimental data for the line complex observed at $E_g$ = 4.44 MeV in irradiations of the two targets. Finally, we emphasize the astrophysical implications of our results.
△ Less
Submitted 18 February, 2023; v1 submitted 25 April, 2022;
originally announced April 2022.
-
Stop** powers of LiF thin films deposited onto self-supporting Al foils for swift protons
Authors:
Smail Damache,
Djamel Moussa,
Saad Ouichaoui
Abstract:
The energy losses of ~(0.273-3.334) MeV protons in Lithium Fluoride thin films deposited by vacuum evaporation onto self-supporting Al foils have been measured using the transmission method. The thicknesses of selected and used Lithium Fluoride/Al target samples were accurately determined via systematic energy loss measurements for alpha particles from a very thin mixed 241Am/239Pu/233U radioactiv…
▽ More
The energy losses of ~(0.273-3.334) MeV protons in Lithium Fluoride thin films deposited by vacuum evaporation onto self-supporting Al foils have been measured using the transmission method. The thicknesses of selected and used Lithium Fluoride/Al target samples were accurately determined via systematic energy loss measurements for alpha particles from a very thin mixed 241Am/239Pu/233U radioactive source. The samples were investigated in detail for their stoichiometry and their impurity contents by backscattering Rutherford spectrometry and nuclear reaction analysis. Then, Lithium Fluoride stop** powers have been determined with overall relative uncertainty of less than 2.7% arising mainly from errors in the determination of target sample thicknesses. These S(E) data are reported and discussed in comparison to previous experimental data sets from the literature and to values calculated by the Sigmund-Schinner binary collision stop** theory both for molecular Lithium Fluoride and for the Lithium Fluoride compound assuming the Bragg's additivity rule. Our S(E) data show to be in excellent agreement with the latter theory for molecular Lithium Fluoride over the whole proton energy range explored, which supports the use of modified hydrogenic wave functions for evaluating atomic shell corrections in the case of low-Z2 target materials. In contrast, they exhibit a slightly increasing deviation from theoretical values derived for the Lithium Fluoride compound with assuming stop** force additivity as the proton energy decreases from ~400 Kev towards lower proton velocities.
△ Less
Submitted 21 May, 2021;
originally announced May 2021.
-
Measurement and analysis of nuclear $γ$-ray production cross sections in proton interactions with Mg, Si and Fe nuclei abundant in astrophysical sites over the incident energy range $E=30-66$ MeV
Authors:
W. Yahia-Cherif,
S. Ouichaoui,
J. Kiener,
E. A. Lawrie,
J. J. Lawrie,
V. Tatischeff,
A. Belhout,
D. Moussa,
P. Papka,
H. Benhabiles,
T. D. Bucher,
A. Chafa,
J. L. Conradie,
S. Damache,
M. Debabi,
I. Deloncle,
J. L. Easton,
C. Hamadache,
F. Hammache,
P. Jones,
B. V. Kheswa,
N. A. Khumalo,
T. Lamula,
S. N. T. Majola,
J. Ndayishimye
, et al. (7 additional authors not shown)
Abstract:
Gamma-ray production cross section excitation functions have been measured for $30$, $42$, $54$ and $66$ MeV proton beams accelerated onto targets of astrophysical interest, $^{nat}$C, C + O (Mylar), $^{nat}$Mg, $^{nat}$Si and $^{56}$Fe, at the Sector Separated Cyclotron (SSC) of iThemba LABS (near Cape Town, South Africa). The AFRODITE array equipped with 8 Compton suppressed HPGe clover detector…
▽ More
Gamma-ray production cross section excitation functions have been measured for $30$, $42$, $54$ and $66$ MeV proton beams accelerated onto targets of astrophysical interest, $^{nat}$C, C + O (Mylar), $^{nat}$Mg, $^{nat}$Si and $^{56}$Fe, at the Sector Separated Cyclotron (SSC) of iThemba LABS (near Cape Town, South Africa). The AFRODITE array equipped with 8 Compton suppressed HPGe clover detectors was used to record $γ$-ray data. For known, intense $γ$-ray lines the previously reported experimental data measured up to $E_{p}\simeq$ $25$ MeV at the Washington and Orsay tandem accelerators were extended to higher proton energies. Our experimental data for the last 3 targets are reported here and discussed with respect to previous data and the Murphy \textit{et al.} compilation [ApJS 183, 142 (2009)], as well as to predictions of the nuclear reaction code TALYS. The overall agreement between theory and experiment obtained in first-approach calculations using default input parameters of TALYS has been appreciably improved by using modified optical model potential (OMP), deformation, and level density parameters. The OMP parameters have been extracted from theoretical fits to available experimental elastic/inelastic nucleon scattering angular distribution data by means of the coupled-channels reaction code OPTMAN. Experimental data for several new $γ$-ray lines are also reported and discussed. The astrophysical implications of our results are emphasised.
△ Less
Submitted 20 January, 2020;
originally announced January 2020.
-
Strongly Consistent of Kullback-Leibler Divergence Estimator and Tests for Model Selection Based on a Bias Reduced Kernel Density Estimator
Authors:
Papa Ngom,
Freedath Djibril Moussa,
Jean de Dieu Nkurunziza
Abstract:
In this paper, we study the strong consistency of a bias reduced kernel density estimator and derive a strongly con- sistent Kullback-Leibler divergence (KLD) estimator. As application, we formulate a goodness-of-fit test and an asymptotically standard normal test for model selection. The Monte Carlo simulation show the effectiveness of the proposed estimation methods and statistical tests.
In this paper, we study the strong consistency of a bias reduced kernel density estimator and derive a strongly con- sistent Kullback-Leibler divergence (KLD) estimator. As application, we formulate a goodness-of-fit test and an asymptotically standard normal test for model selection. The Monte Carlo simulation show the effectiveness of the proposed estimation methods and statistical tests.
△ Less
Submitted 18 May, 2018;
originally announced May 2018.
-
Stop** powers and energy loss straggling for MeV protons in a kapton polyimide thin film
Authors:
S. Damache,
S. Djaroum,
S. Ouichaoui,
L. Amari,
D. Moussa
Abstract:
The energy loss and energy loss straggling widths have been measured in transmission for MeV protons traversing a thin kapton polyimide foil. In a prior step, the thickness and non-uniformity of the target foil were carefully investigated. The overall relative uncertainties in the stop** power and energy loss straggling variance data amount, respectively, to less than 2% and 8%. The experimental…
▽ More
The energy loss and energy loss straggling widths have been measured in transmission for MeV protons traversing a thin kapton polyimide foil. In a prior step, the thickness and non-uniformity of the target foil were carefully investigated. The overall relative uncertainties in the stop** power and energy loss straggling variance data amount, respectively, to less than 2% and 8%. The experimental data show to be in excellent agreement with available previous ones and with those compiled in the ICRU-49 report. They are fully consistent with the predictions of Sigmund-Schinner's binary collision theory of electronic stop** over the whole proton energy range explored. An average deviation of ~2.5% relative to values calculated by the SRIM-2008 code, likely due to effects of valence electrons involving the , and bonds, is however observed at low proton velocities. The measured energy loss straggling data, which are unique to our knowledge, are found to be in good agreement with values derived by the classical Bohr formula for keV but they significantly exceed Bohr's collisional energy loss straggling at lower proton velocities where target electrons can no longer be considered as free. They also show to be consistent with the predictions of the Bethe-Livingston and Sigmund-Schinner theories over the low proton velocity region ( keV). However, they are significantly overestimated by these theories over the intermediate and high proton velocity regions, which may be due to bunching effect by inner shell electrons of the polymer target. Besides, our energy loss straggling data are in better overall consistency with the Yang, O'Connor and Wang empirical formula for keV, while deviations above the latter amounting up to ~18% are observed at lower proton velocities.
Keywords: Swift protons; Kapton polyimide; Stop** power; Energy loss straggling; Binary collision theory.
△ Less
Submitted 18 September, 2017;
originally announced September 2017.
-
Gamma ray production cross sections in proton induced reactions on natural Mg, Si and Fe targets over the proton energy range 30 up to 66 MeV
Authors:
W. Yahia-Chérif,
S. Ouichaoui,
J. Kiener,
V. Tatischeff,
E. Lawrie,
J. J. Lawrie,
A. Belhout,
H. Benhabiles,
T. D. Bucher,
A. Chafa,
S. Damache,
M. Debabi,
I. Deloncle,
J. L. Easton,
C. Hamadache,
F. Hammache,
P. Jones,
B. V. Kheswa,
N. Khumalo,
T. Lamula,
S. T. H. Majola,
D. Negi,
J. Ndayishimye,
S. P. Noncolela,
D. Moussa
, et al. (7 additional authors not shown)
Abstract:
Gamma-ray excitation functions have been measured for 30, 42, 54 and 66 MeV proton beams accelerated onto C + O (Mylar), Mg, Si, and Fe targets of astrophysical interest at the separate-sector cyclotron of iThemba LABS in Somerset West (Cape Town, South Africa). A large solid angle, high energy resolution detection system of the Eurogam type was used to record Gamma-ray energy spectra. Derived pre…
▽ More
Gamma-ray excitation functions have been measured for 30, 42, 54 and 66 MeV proton beams accelerated onto C + O (Mylar), Mg, Si, and Fe targets of astrophysical interest at the separate-sector cyclotron of iThemba LABS in Somerset West (Cape Town, South Africa). A large solid angle, high energy resolution detection system of the Eurogam type was used to record Gamma-ray energy spectra. Derived preliminary results of Gamma-ray line production cross sections for the Mg, Si and Fe target nuclei are reported and discussed. The current cross section data for known, intense Gamma-ray lines from these nuclei consistently extend to higher proton energies previous experimental data measured up to Ep ~ 25 MeV at the Orsay and Washington tandem accelerators. Data for new Gamma-ray lines observed for the first time in this work are also reported.
△ Less
Submitted 9 July, 2015;
originally announced July 2015.
