The design and characterization of a low-cost, open-source auditory delivery system to deliver high performance auditory stimuli is presented. The system includes a high-fidelity sound card and audio amplifier devices with low-latency and wide bandwidth targeted for behavioral neuroscience research. The characterization of the individual devices and the entire system is performed, providing a thorough audio characterization data for varying frequencies and sound levels. The system implements the open-source Harp protocol, enabling the hardware timestamping of devices and seamless synchronization with other Harp devices.

BACKGROUND: Heart failure (HF) contributes greatly to morbidity, mortality, and health care costs worldwide. Hospital readmission rates are tracked closely and determine federal reimbursement dollars. No current modality or technology allows for accurate measurement of relevant HF parameters in ambulatory, rural, or underserved settings. This limits the use of telehealth to diagnose or monitor HF in ambulatory patients.
OBJECTIVE: This study describes a novel HF diagnostic technology using audio recordings from a standard mobile phone.
METHODS: This prospective study of acoustic microphone recordings enrolled convenience samples of patients from 2 different clinical sites in 2 separate areas of the United States. Recordings were obtained at the aortic (second intercostal) site with the patient sitting upright. The team used recordings to create predictive algorithms using physics-based (not neural networks) models. The analysis matched mobile phone acoustic data to ejection fraction (EF) and stroke volume (SV) as evaluated by echocardiograms. Using the physics-based approach to determine features eliminates the need for neural networks and overfitting strategies entirely, potentially offering advantages in data efficiency, model stability, regulatory visibility, and physical insightfulness.
RESULTS: Recordings were obtained from 113 participants. No recordings were excluded due to background noise or for any other reason. Participants had diverse racial backgrounds and body surface areas. Reliable echocardiogram data were available for EF from 113 patients and for SV from 65 patients. The mean age of the EF cohort was 66.3 (SD 13.3) years, with female patients comprising 38.3% (43/113) of the group. Using an EF cutoff of ≤40% versus >40%, the model (using 4 features) had an area under the receiver operating curve (AUROC) of 0.955, sensitivity of 0.952, specificity of 0.958, and accuracy of 0.956. The mean age of the SV cohort was 65.5 (SD 12.7) years, with female patients comprising 34% (38/65) of the group. Using a clinically relevant SV cutoff of <50 mL versus >50 mL, the model (using 3 features) had an AUROC of 0.922, sensitivity of 1.000, specificity of 0.844, and accuracy of 0.923. Acoustics frequencies associated with SV were observed to be higher than those associated with EF and, therefore, were less likely to pass through the tissue without distortion.
CONCLUSIONS: This work describes the use of mobile phone auscultation recordings obtained with unaltered cellular microphones. The analysis reproduced the estimates of EF and SV with impressive accuracy. This technology will be further developed into a mobile app that could bring screening and monitoring of HF to several clinical settings, such as home or telehealth, rural, remote, and underserved areas across the globe. This would bring high-quality diagnostic methods to patients with HF using equipment they already own and in situations where no other diagnostic and monitoring options exist.

Musical depth, which encompasses the intellectual and emotional complexity of music, is a robust dimension that influences music preference. However, there remains a dearth of research exploring the relationship between lyrics and musical depth. This study addressed this gap by analyzing linguistic inquiry and word count-based lyric features extracted from a comprehensive dataset of 2372 Chinese songs. Correlation analysis and machine learning techniques revealed compelling connections between musical depth and various lyric features, such as the usage frequency of emotion words, time words, and insight words. To further investigate these relationships, prediction models for musical depth were constructed using a combination of audio and lyric features as inputs. The results demonstrated that the random forest regressions (RFR) that integrated both audio and lyric features yielded superior prediction performance compared to those relying solely on lyric inputs. Notably, when assessing the feature importance to interpret the RFR models, it became evident that audio features played a decisive role in predicting musical depth. This finding highlights the paramount significance of melody over lyrics in effectively conveying the intricacies of musical depth.

Identification of different species of animals has become an important issue in biology and ecology. Ornithology has made alliances with other disciplines in order to establish a set of methods that play an important role in the birds\' protection and the evaluation of the environmental quality of different ecosystems. In this case, the use of machine learning and deep learning techniques has produced big progress in birdsong identification. To make an approach from AI-IoT, we have used different approaches based on image feature comparison (through CNNs trained with Imagenet weights, such as EfficientNet or MobileNet) using the feature spectrogram for the birdsong, but also the use of the deep CNN (DCNN) has shown good performance for birdsong classification for reduction of the model size. A 5G IoT-based system for raw audio gathering has been developed, and different CNNs have been tested for bird identification from audio recordings. This comparison shows that Imagenet-weighted CNN shows a relatively high performance for most species, achieving 75% accuracy. However, this network contains a large number of parameters, leading to a less energy efficient inference. We have designed two DCNNs to reduce the amount of parameters, to keep the accuracy at a certain level, and to allow their integration into a small board computer (SBC) or a microcontroller unit (MCU).

BACKGROUND: The digital era has witnessed an escalating dependence on digital platforms for news and information, coupled with the advent of \"deepfake\" technology. Deepfakes, leveraging deep learning models on extensive data sets of voice recordings and images, pose substantial threats to media authenticity, potentially leading to unethical misuse such as impersonation and the dissemination of false information.
OBJECTIVE: To counteract this challenge, this study aims to introduce the concept of innate biological processes to discern between authentic human voices and cloned voices. We propose that the presence or absence of certain perceptual features, such as pauses in speech, can effectively distinguish between cloned and authentic audio.
METHODS: A total of 49 adult participants representing diverse ethnic backgrounds and accents were recruited. Each participant contributed voice samples for the training of up to 3 distinct voice cloning text-to-speech models and 3 control paragraphs. Subsequently, the cloning models generated synthetic versions of the control paragraphs, resulting in a data set consisting of up to 9 cloned audio samples and 3 control samples per participant. We analyzed the speech pauses caused by biological actions such as respiration, swallowing, and cognitive processes. Five audio features corresponding to speech pause profiles were calculated. Differences between authentic and cloned audio for these features were assessed, and 5 classical machine learning algorithms were implemented using these features to create a prediction model. The generalization capability of the optimal model was evaluated through testing on unseen data, incorporating a model-naive generator, a model-naive paragraph, and model-naive participants.
RESULTS: Cloned audio exhibited significantly increased time between pauses (P<.001), decreased variation in speech segment length (P=.003), increased overall proportion of time speaking (P=.04), and decreased rates of micro- and macropauses in speech (both P=.01). Five machine learning models were implemented using these features, with the AdaBoost model demonstrating the highest performance, achieving a 5-fold cross-validation balanced accuracy of 0.81 (SD 0.05). Other models included support vector machine (balanced accuracy 0.79, SD 0.03), random forest (balanced accuracy 0.78, SD 0.04), logistic regression, and decision tree (balanced accuracies 0.76, SD 0.10 and 0.72, SD 0.06). When evaluating the optimal AdaBoost model, it achieved an overall test accuracy of 0.79 when predicting unseen data.
CONCLUSIONS: The incorporation of perceptual, biological features into machine learning models demonstrates promising results in distinguishing between authentic human voices and cloned audio.

BACKGROUND: Vocal biomarkers, derived from acoustic analysis of vocal characteristics, offer noninvasive avenues for medical screening, diagnostics, and monitoring. Previous research demonstrated the feasibility of predicting type 2 diabetes mellitus through acoustic analysis of smartphone-recorded speech. Building upon this work, this study explores the impact of audio data compression on acoustic vocal biomarker development, which is critical for broader applicability in health care.
OBJECTIVE: The objective of this research is to analyze how common audio compression algorithms (MP3, M4A, and WMA) applied by 3 different conversion tools at 2 bitrates affect features crucial for vocal biomarker detection.
METHODS: The impact of audio data compression on acoustic vocal biomarker development was investigated using uncompressed voice samples converted into MP3, M4A, and WMA formats at 2 bitrates (320 and 128 kbps) with MediaHuman (MH) Audio Converter, WonderShare (WS) UniConverter, and Fast Forward Moving Picture Experts Group (FFmpeg). The data set comprised recordings from 505 participants, totaling 17,298 audio files, collected using a smartphone. Participants recorded a fixed English sentence up to 6 times daily for up to 14 days. Feature extraction, including pitch, jitter, intensity, and Mel-frequency cepstral coefficients (MFCCs), was conducted using Python and Parselmouth. The Wilcoxon signed rank test and the Bonferroni correction for multiple comparisons were used for statistical analysis.
RESULTS: In this study, 36,970 audio files were initially recorded from 505 participants, with 17,298 recordings meeting the fixed sentence criteria after screening. Differences between the audio conversion software, MH, WS, and FFmpeg, were notable, impacting compression outcomes such as constant or variable bitrates. Analysis encompassed diverse data compression formats and a wide array of voice features and MFCCs. Wilcoxon signed rank tests yielded P values, with those below the Bonferroni-corrected significance level indicating significant alterations due to compression. The results indicated feature-specific impacts of compression across formats and bitrates. MH-converted files exhibited greater resilience compared to WS-converted files. Bitrate also influenced feature stability, with 38 cases affected uniquely by a single bitrate. Notably, voice features showed greater stability than MFCCs across conversion methods.
CONCLUSIONS: Compression effects were found to be feature specific, with MH and FFmpeg showing greater resilience. Some features were consistently affected, emphasizing the importance of understanding feature resilience for diagnostic applications. Considering the implementation of vocal biomarkers in health care, finding features that remain consistent through compression for data storage or transmission purposes is valuable. Focused on specific features and formats, future research could broaden the scope to include diverse features, real-time compression algorithms, and various recording methods. This study enhances our understanding of audio compression\'s influence on voice features and MFCCs, providing insights for developing applications across fields. The research underscores the significance of feature stability in working with compressed audio data, laying a foundation for informed voice data use in evolving technological landscapes.

Research in machine learning (ML) algorithms using natural behavior (i.e., text, audio, and video data) suggests that these techniques could contribute to personalization in psychology and psychiatry. However, a systematic review of the current state of the art is missing. Moreover, individual studies often target ML experts who may overlook potential clinical implications of their findings. In a narrative accessible to mental health professionals, we present a systematic review conducted in 5 psychology and 2 computer science databases. We included 128 studies that assessed the predictive power of ML algorithms using text, audio, and/or video data in the prediction of anxiety and posttraumatic stress disorder. Most studies (n = 87) were aimed at predicting anxiety, while the remainder (n = 41) focused on posttraumatic stress disorder. They were mostly published since 2019 in computer science journals and tested algorithms using text (n = 72) as opposed to audio or video. Studies focused mainly on general populations (n = 92) and less on laboratory experiments (n = 23) or clinical populations (n = 13). Methodological quality varied, as did reported metrics of the predictive power, hampering comparison across studies. Two-thirds of studies, which focused on both disorders, reported acceptable to very good predictive power (including high-quality studies only). The results of 33 studies were uninterpretable, mainly due to missing information. Research into ML algorithms using natural behavior is in its infancy but shows potential to contribute to diagnostics of mental disorders, such as anxiety and posttraumatic stress disorder, in the future if standardization of methods, reporting of results, and research in clinical populations are improved.

BACKGROUND: The maintenance of oral hygiene and subsequent health related issues in visually handicapped children is a challenging task. Hence, tools must be used to ensure good oral health in these children. The study aimed to analyze the effectiveness of preventive programs on oral health using specialized audio and tactile aids in visually impaired school children.
METHODS: 100 visually handicapped children were selected for the study. They were divided into two groups: Group A (Training using Braille) and Group B (training by means of audio aids). Children were trained using appropriate tools and oral health was assessed using Plaque and Gingival bleeding indices.
METHODS: Independent \'T-test\' was used for comparing mean ± SD values.
RESULTS: Statistically significant improvements in both plaque and gingival bleeding indices were obtained on follow-up observations at 3rd and 6th months.
CONCLUSIONS: The use of specialized tactile and audio tools significantly improved the oral health status of visually impaired school children.

Whilst radio, podcasts, and music streaming are considered unique audio formats that offer brands different opportunities, limited research has explored this notion. This current study analyses how the brain responds to these formats and suggests that they offer different branding opportunities. Participants\' engagement, attitude, attention, memory, and physiological arousal were measured while each audio format was consumed. The results revealed that music streaming elicited more positive attitudes, higher attention, greater levels of memory encoding, and increased physiological arousal compared to either radio or podcasts. This study emphasises the importance for brands of utilising diverse audio channels for unique branding and marketing opportunities.

BACKGROUND: In the global strategy to eliminate leprosy, there remains a need for early case detection to successfully interrupt transmissions. Poor knowledge about leprosy and leprosy-related stigma are key drivers of delayed diagnosis and treatment. Sensitization campaigns to inform and increase awareness among the general population are an integral part of many national neglected tropical disease programs. Despite their importance, the effectiveness of such campaigns has not been rigorously studied in the West African context. A multilingual rural setting with low health literacy in this region presents challenges to the potential impact of sensitization campaigns.
OBJECTIVE: The primary objective of this study is to assess the causal effect of common practice community sensitization campaigns on leprosy-related knowledge and stigma at the community level and among community health volunteers. Additionally, we will test the potential of novel educational audio tools in the 15 most prominent local languages to overcome literacy and language barriers and amplify sensitization campaigns.
METHODS: We will conduct a cluster randomized controlled trial using a sequential mixed methods approach in 60 rural communities across all regions of Togo, West Africa. The study features 2 intervention arms and 1 control arm, with intervention and control assignments made at the community level through randomization. Communities in intervention arm 1 will receive a sensitization campaign in line with the current Togolese national neglected tropical disease program. Communities in intervention arm 2 will receive the same sensitization campaign along with educational audio tools distributed to community households. The control arm will receive no intervention before data collection. Quantitative outcome measures on knowledge and stigma will be collected from a random sample of 1200 individuals. Knowledge will be assessed using the 9-item standardized Knowledge, Attitudes, and Practices Questionnaire. Stigma will be measured using the 7-item Social Distance Scale and the 15-item Explanatory Model Interview Catalogue Community Stigma Scale. We will estimate intention-to-treat effects at the individual level, comparing the outcomes of the intervention and control arms. In an accompanying qualitative component, we will conduct in-depth interviews with community members, community health volunteers, and health care workers in both treatment arms and the control arm to explore intervention and stigma-related experiences.
RESULTS: This paper describes and discusses the protocol for a mixed methods cluster randomized controlled trial. Data collection is planned to be completed in June 2024, with ongoing data analysis. The first results are expected to be submitted for publication by the end of 2024.
CONCLUSIONS: This trial will be among the first to test the causal effectiveness of community-based sensitization campaigns and audio tools to increase knowledge and reduce leprosy-related stigma. As such, the results will inform health policy makers, decision-makers, and public health practitioners designing sensitization campaigns in rural multilingual settings.
BACKGROUND: German Clinical Trials Register DRKS00029355; https://drks.de/search/en/trial/DRKS00029355.
UNASSIGNED: DERR1-10.2196/52106.