Effects of language mismatch in automatic forensic voice comparison using deep learning embeddings

In forensic voice comparison, deep learning has become widely popular recently. It is mainly used to learn speaker representations, called embeddings or embedding vectors. Speaker embeddings are often trained using corpora mostly containing widely spoken languages. Thus, language dependency is an important factor in automatic forensic voice comparison, especially when the target language is linguistically very different from that the model is trained on. In the case of a low-resource language, developing a corpus for forensic purposes containing enough speakers to train deep learning models is costly. This study aims to investigate whether a model pre-trained on multilingual (mostly English) corpus can be used on a target low-resource language (here, Hungarian), not represented by the model. Often multiple samples are not available from the offender (unknown speaker). Samples are therefore compared pairwise with and without speaker enrollment for suspect (known) speakers. Two corpora are used that were developed especially for forensic purposes and a third that is meant for traditional speaker verification. Speaker embedding vectors are extracted by the x-vector and ECAPA-TDNN techniques. Speaker verification was evaluated in the likelihood-ratio framework. A comparison is made between the language combinations (modeling, LR calibration, and evaluation). The results were evaluated by Cllr_min and EER metrics. It was found that the model pre-trained on a different language but on a corpus with a significant number of speakers can be used on samples with language mismatch. Sample duration and speaking style also seem to affect the performance.     

Lead isotope measurement of primer gunshot residues and likelihood ratio predictions for forensic cartridge discrimination and individualization in China

Gunshot residues (GSR), cartridge projectiles, and casings are frequently encountered evidence in gun-related forensic investigations. However, in circumstances where the investigation of striation marks is impossible, such as unrecovered or deformed projectiles and cartridge casings, GSR deposited on the hands or clothes of the shooter and victim-related items can provide information to establish a link between the suspect, the firearms used, and the victim. Since the formula of primers used by all cartridge manufacturers in China is identical, links based on the conventional morphological and compositional analysis of GSR are difficult to establish. However, the abundance of lead isotopes in primer components of lead styphnate varies significantly, and a fundamental understanding of these differences may facilitate the validation of primer (p)GSR evidence in forensic investigations. Here, 44 pGSR samples were characterized by Pb isotope ratios of ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, and ²⁰⁸Pb/²⁰⁴Pb using laser ablation multicollector inductively coupled plasma mass spectrometry. There was no obvious mass fractionation of the lead isotope ratios of the primers from individual cartridges analyzed before and after the shooting process, thereby establishing a basis for the comparison of pGSR and unfired cartridges. Evaluation of the results using univariate likelihood ratio (LR) computations revealed low rates of misleading evidence (<0.53%) The results demonstrated that lead isotope ratio analysis of pGSR and LR predictions can provide a practicable method for forensic cartridge discrimination and individualization.     

ForensicNet: Modern convolutional neural network-based image forgery detection network

The advancements in Image editing techniques produce realistic-looking artificial images with ease. These images can easily circumvent the forensic systems making the authentication process more tedious and difficult. To overcome this problem, we introduce a modern convolutional neural network (CNN) named ForensicNet, inspired by the recent developments in computer vision. The three major contributions of our CNNs are inverted bottleneck, separate downsampling layers, and using depth-wise convolutions for mixing information in the spatial dimension. The inverted bottlenecks help improve accuracy and reduce network parameters/FLOPs. The separate downsampling layers help converge the network. The normalization layers also help stabilize training whenever the spatial resolution is changed. The depth-wise convolution is a grouped convolution where the number of groups and channels are the same. The experiments show that ForensicNet outperforms the state-of-the-art methods by a large margin.     

Increased lung weight in fatal intoxications is not unique to opioid drugs

Fatal intoxications with opioids are known to be associated with an increased lung weight, as well as with brain and pulmonary edema and urinary retention. However, there is evidence to suggest that fatal intoxications with non-opioid substances are also associated with increased lung weight; however, the latter aspect has not been comprehensively analyzed. To determine to what extent opioid and non-opioid substances are associated with increased lung and brain weight, we studied these organs in cases where the cause of death was attributed to intoxication with a single agent. Using data from cases autopsied at the National Board of Forensic Medicine (NBFM) in Sweden from 2009 through 2019 where the cause of death was attributed to a single substance, we created models of combined lung weight and brain weight. The models used age and sex as predictors as well as nested varying effects for the specific intoxicant and category of intoxicant. Suicidal hanging with negative toxicology cases served as controls. The population majority was male among both intoxications (68%) and controls (83%). The most common single substance group was opioids. All tested substances were associated with heavier lungs than controls, with the largest effect in the opioid group. Our findings show that several substances are associated with increased lung weight and that among intoxication deaths there is no difference in expected brain weight between substances. Hence, heavy lungs, without a reasonable explanation, should prompt a broad toxicological screening.     

An analysis of data curated from 5 years of identifying human remains

An archive of 5 years of cases involving the identification of human remains was curated, collecting information on: The sample type submitted, the number of STR loci yielding interpretable results, the kinship challenge posed, and the outcome for the case. A total of 129 cases of remains ID were investigated using manual DNA extraction and recovery methods with amplification of STR markers using the Power Plex 21 multiplex STR kit from Promega Corp. In 52 cases, blood spots collected by the ME were provided as sample and in 100% of those cases, probabilities of relatedness to the reference samples was ≥99%. In 77 cases, tissue other than blood was provided as a source of DNA. These other samples were grouped categorically into long bones (femur and tibia; 40 cases), skull bones/teeth (11 cases), other bones (16 cases), and tissue (normally adherent to bone) (10 cases). Reference samples provided for cases included alleged parents or child(ren) of the victim (86 cases), alleged full siblings of the victim (38 cases), or alleged second-order relatives (five cases). The overall success rate in confirming the identity of the source of the remains in these cases was 89.2%. Our results demonstrate that a laboratory can be often successful identifying human remains using methods easily implemented in any DNA typing laboratory.     

Suicide of an adolescent girl with sodium nitrite ordered on the internet

Nitrites are commonly used in the chemical, pharmaceutical, and food industries. Recently, they have been identified in cases of voluntary intoxication. We report the case of a 13-year-old girl who was found lifeless on her bed next to a glass containing a white powder and a bottle containing a white powder with a moistened appearance. External examination and autopsy revealed a nonspecific asphyxia syndrome, which was confirmed by the pathological analysis. Analysis of the samples revealed metoclopramide in the peripheral blood at a concentration of 0.402 mg/L (LC–HRMS). An analysis of the gastric contents was carried out after sodium nitrite was detected in the powders found near the body (Raman spectrometry). Nitrites were found in the gastric fluid at a concentration of 30.9 mg/L. Death occurred secondary to anoxia, following ingestion of nitrites; suicide kits are available on the web and nitrites are relatively easy to source and inexpensive. Nitrites are delivered in powder form to be dissolved in liquid, which may then be consumed with metoclopramide (or an alternative anti-emetic drug) to maximize absorption and reduce emesis. The toxic effect of nitrites lies in their oxidizing power, causing the transformation of hemoglobin into methemoglobin, which, when it accumulates, induces tissue anoxia resulting in death. There has been an alarming increase in the number of cases linked to suicide using nitrites or a nitrite suicide kit. The fact that nitrites are readily available online underscores the importance of establishing effective preventive measures such as limiting the access and use of this chemical.     

Identification of document paper using hybrid feature extraction

Document forgery is a significant issue in Korea, with around ten thousand cases reported every year. Analyzing paper plays a crucial role in examining questionable documents such as marketable securities and contracts, which can aid in solving criminal cases of document forgery. Paper analysis can also provide essential insights in other types of criminal cases, serving as an important clue for solving cases such as the source of a blackmail letter. The papermaking process generates distinct forming fabric marks and formations, which are critical features for paper classification. These characteristics are observable under transmitted light and are created by the forming fabric pattern and the distribution of pulp fibers, respectively. In this study, we propose a novel approach for paper identification based on hybrid features. This method combines texture features extracted from images converted using the gray-level co-occurrence matrix (GLCM) approach and a convolutional neural network (CNN), with another set of features extracted by the CNN using the same images as input. We applied the proposed method to classification tasks for seven major paper brands available in the Korean market, achieving an accuracy of 97.66%. The results confirm the applicability of this method for visually inspecting paper products and demonstrate its potential for assisting in solving criminal cases involving document forgery.     

Voltammetric analysis of luminescent markers in gunshot residues

Currently, SEM-EDS is used to detect gunshot residue (GSR) from the presence of Ba, Pb, and Sb in the sample. However, the development of new nontoxic ammunition (NTA) has prevented conventional metals from being found. In this work, we aim to determine the presence of an inorganic luminescent chemical marker based on rare earth in gunshot residues using the technique of squarewave voltammetry (SWV). After firing, the luminescent complex [(Eu₂Zr)(btc)₃(Hbtc)_0.5.6H₂O], which is used as a chemical marker, can be detected under a UV lamp. An aqueous solution with 0.1 mol L⁻¹ KCl as supporting electrolyte can be easily collected on carbon paste electrode surfaces for SWV analysis A = 100 mV, f = 10 Hz, and step potential of 5 mV are required. The luminescent marker incorporated into the carbon paste electrode showed two anodic peak currents in the region of 0.4 V (vs Ag/AgCl) and at 0.75 V (vs Ag/AgCl) and also a cathodic one in 0.4 V (vs Ag/AgCl). SEM-EDS was able to analyze the same voltammetric results for conventional and nontoxic ammunition containing the luminescent marker. Therefore, voltammetry and SEM-EDS are valid for detecting the new residue marker in GSR. Despite this, the electrochemical method is still more advantageous because of its low cost and lack of expensive equipment and supplies in forensic laboratories.     

Rapid and non-destructive approach for characterization and differentiation of sealing wax using ATR-FTIR spectroscopy

Sealing wax is used for maintaining the integrity and authenticity of a document or physical evidence. Any tampering with the seal calls into question the overall integrity and authenticity of the tangible evidence or document. In these circumstances, determining the authenticity of the sealing material (physical and chemical) becomes imperative. In this study, ATR-FTIR spectroscopy supported by chemometrics has been used to differentiate sealing wax samples belonging to 12 different brands available across India. All the samples were first melted, cooled, and then analyzed using ATR-FTIR spectroscopy in the mid-infrared region (4000–600 cm⁻¹). The obtained spectra were first examined visually for the presence of different functional groups. Principal component analysis (PCA) and principal component analysis-linear discriminant analysis (PCA-LDA) were employed to analyze the sample clustering patterns and to categorize them into their respective groups, respectively. For classification, a PCA-LDA training model was applied, and it demonstrated 95.83% accuracy. The validation test resulted in an accuracy of 83.33%. PCA-LDA model offered 100% accurate prediction for samples on various substrates, including cloth, cardboard, and paper. A blind study was also performed using five unknown samples, which were accurately classified into their respective groups. PCA-LDA model will be helpful in providing investigative leads by linking a questioned sealing wax sample with its respective group.     

The rise and rise of fentanyl in postmortem casework

Forensic toxicology laboratories are navigating a period of time with increasing drug overdose deaths, an opioid epidemic, the impact of the COVID-19 pandemic, and the illicit drug market flooded with novel psychoactive substances. In New York City, the Department of Forensic Toxicology has experienced a 56% increase in postmortem casework in the past decade with fentanyl detected in 80% of all overdose deaths. Over a period of 2.5 years, 15,638 postmortem cases were tested for the presence of fentanyl and fentanyl analogs using liquid-chromatography tandem mass spectrometry (LCMSMS). Fentanyl was detected in approximately one third of cases and of these 4447 cases with femoral blood. A twofold increase in cases with high concentrations of fentanyl (>100 ng/mL) was observed between 2021 and 2022. The minor metabolite and precursor chemical, 4-ANPP (4-anilino-N-phenethylpiperidine) may help differentiate between illicit and licit fentanyl. 4-ANPP blood concentrations were <10 ng/mL in 98% of the cases and the 4-ANPP:fentanyl ratio was <0.67 for 99.1% of blood specimens. Only six cases had 4-ANPP concentrations higher than the corresponding fentanyl blood concentration. This study also highlights, the changing fentanyl analogs found in postmortem cases since 2016 in NYC with the emergence of fluorofentanyl initially identified in 2020 and continuing to dominate in comparison with the prevalence of other analogs, many of which are no longer detected in casework. The detection of one of the latest drugs to be mixed with fentanyl, namely xylazine, has also increased in prevalence by 36.7% in 2022 compared with 2021.