SEARCH · Search NASA

Results for “Speech analysis”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

11 records

All sources NASA NTRS DOE OSTI

DOE OSTI2025

A Model Based Approach to Extract Health Information from Textual Data

In current nuclear power plants (NPPs) a large amount of condition-based data is being generated and stored to assess and monitor component health and performance. The format of this data can be either numeric (e.g., pump vibration data) or textual (e.g., condition report which assess component health). While assessing component health from numeric data can be performed with a large variety of methods, the extraction of information from textual data still remains a challenge. Natural language processing (NLP) methods are starting to be deployed in current NPPs mainly to filter out incident reports (IRs) that are not safety related by employing supervised machine learning methods. However, these methods do not really provide the quantitative information that might be contained in IRs. This paper presents an approach to extract information from textual data (e.g., from IRs, maintenance reports) that is based on NLP data analytics methods coupled with model-based system engineer (MBSE) models. NLP methods are employed to perform syntactic and semantic analyses. Syntactic analysis analyzes the grammatical structure of a sentence; such analysis includes: part of speech (POS) tagging (i.e., identification of grammatic elements of each string - e.g., nouns, verbs), named entity recognition (i.e., identification of text entities - e.g., names, dates, events), and relation extraction (e.g., coreference resolution). On the other hand, semantic analysis is designed to analyze the logic structure of a sentence. Through a specific set of rules, our methods can identify whether a sentence contains health information of a component (e.g., degraded performance, anomaly behavior) or the causal relationship between two events (i.e., a cause-effect pair). An innovative element of our approach is that semantic analysis relies on MBSE models to identify links between textual elements. MBSE are diagrams designed to represent system and component dependencies (from both a form and functional point of view). In our approach, MBSE models emulate system engineer knowledge about component/system architecture. This paper presents in detail how the integration of NLP methods and MBSE models is performed. Few analysis examples focusing on centrifugal pumps are presented.

97 - MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Measuring Equality in Machine Learning Security Defenses: A Case Study in Speech Recognition

Over the past decade, the machine learning security community has developed a myriad of defenses for evasion attacks. An understudied question in that community is: for whom do these defenses defend? This work considers common approaches to defending learned systems and how security defenses result in performance inequities across different sub-populations. We outline appropriate parity metrics for analysis and begin to answer this question through empirical results of the fairness implications of machine learning security methods. We find that many methods that have been proposed can cause direct harm, like false rejection and unequal benefits from robustness training. The framework we propose for measuring defense equality can be applied to robustly trained models, preprocessing-based defenses, and rejection methods. We identify a set of datasets with a user-centered application and a reasonable computational cost suitable for case studies in measuring the equality of defenses. In our case study of speech command recognition, we show how such adversarial training and augmentation have non-equal but complex protections for social subgroups across gender, accent, and age in relation to user coverage. We present a comparison of equality between two rejection-based defenses: randomized smoothing and neural rejection, finding randomized smoothing more equitable due to the sampling mechanism for minority groups. This represents the first work examining the disparity in the adversarial robustness in the speech domain and the fairness evaluation of rejection-based defenses.

• Artificial intelligence (AI) / machine learning ↗

DOE OSTI2025

AI foundation models for experimental fusion tasks

Artificial Intelligence (AI) foundation models, while successful in various domains of language, speech, and vision, have not been adopted in production for fusion energy experiments. This brief paper presents how AI foundation models can be used for fusion energy diagnostics, enabling, for example, visual automated logbooks to provide greater insights into chains of plasma events in a discharge, in time for between-shot analysis.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2023

Acoustic scattering by smooth elastic cylinders insonified by directional transceivers: Monostatic theory and experiments

A theoretical model for predicting the acoustic field scattered by an elastic cylinder that is partially insonified by a directional transceiver is proposed in the form of a simple approximate one-dimensional integral. This model accounts for spherical spreading and directivity of the incident waves and extends the formulation used in a preceding article by including effects due to oblique insonification of a long cylinder assuming negligible end-contributions. The scattered field of an infinitely long cylinder for obliquely incident plane waves and point receivers is used to approximate the apparent volume flow of cylinders partially insonified by directional transceivers. The scattered pressure that is derived using the apparent volume flow, in contrast to the previous formulation, is capable of predicting axially propagating guided wave resonances; these natural modes are excited, in addition to circumferential ones, at off-normal incident angles. The model is compared with exact numerical simulations and with previously published as well as new laboratory data. The analysis illustrates the different realistic effects associated with scattering from elastic cylinders insonified by a directional transceiver both theoretically and experimentally.

42 ENGINEERING↗

DOE OSTI2025

Extraction and Analysis of Time Series Data from Building Automation Systems Using Large Language Models

Semantic schemas like Haystack 4, Brick and ASHRAE standard 223 enable the structured, standardized, and machine-readable representation of building data, facilitating interoperability, data integration, and advanced analytics. However, extracting information from these models requires specialized expertise in SPARQL and other programming languages, skills that are not commonly found among building professionals. Recent advancements in Large Language Models (LLMs), such as ChatGPT, enable the construction of queries using natural language, making it easier for individuals to interact with these systems in a manner that resembles everyday speech. However, these methods have not yet been tested on building semantic ontologies. This paper introduces a novel workflow and tool for enabling users to ask questions about a specific building's data, using natural language and receive answers automatically generated by GPT-4o. Our approach integrates semantic ontologies with advanced LLM capabilities to automate three critical steps: (1) generating SPARQL queries to retrieve time series references from ontological models, (2) extracting the corresponding time series data from the Building Automation System, and (3) performing computations and visualizations tailored to the user's query. The proposed method simplifies access to BAS data, allowing both domain experts and non-specialists to conduct sophisticated analyses without needing extensive technical knowledge of semantic web technologies. By demonstrating this pipeline, we facilitate more accessible and scalable data-driven decision-making in building operations and management.

Mulayim, Ozan Baris↗

DOE OSTI2025

Extraction and Analysis of Time Series Data from Building Automation Systems Using Large Language Models

Mulayim, Ozan Baris↗

DOE OSTI2025

Elastomeric Nanocomposite with Solvent‐Free, One Step, In Situ Shear Exfoliation of Graphite to Graphene

A graphene nanoflake (GNF)‐enhanced elastomeric nanocomposite (G‐EMC) is fabricated following an innovative, cost‐effective, single‐step, in situ shear exfoliation (ISE) method from low‐cost bulk material, graphite, where uniform mixing happens simultaneously within the elastomer matrix. Electron microscopy, atomic force microscopy, and photo‐induced force microscopy results show good dispersion of GNFs with exfoliation to a few layers and uniform distribution in the elastomer matrix. X‐ray photoelectron spectroscopy analysis shows less than 1% oxygen‐containing functional groups/impurity, enhanced bonding through the formation of edge sites as fracture occurs across the GNF basal plane, and pi‐pi interactions with newly exfoliated planar basal plane surfaces of the GNFs. Raman spectroscopy results confirm the formation of GNFs with only a few layers of graphene formed by the ISE process. Fabricated 10 wt.% G‐EMC nanocomposites show a 400%–500% increase in strength and fracture toughness. And 35 wt.% G‐EMCs provide an electrical conductivity of 25.64 S m −1 and a sensor gauge factor of 45. The resulting intrinsic piezo resistivity of the fabricated nanocomposite has been exploited to fabricate a multi‐functional wired and wireless sensor for detecting different body movements, speech, human vital functions, solvents, and biomolecules.

36 MATERIALS SCIENCE↗

DOE OSTI2022

Provider Perspectives: Identification and Follow-up of Infants who Are Deaf or Hard of Hearing

Objective Without timely screening, diagnosis, and intervention, hearing loss can cause significant delays in a child's speech, language, social, and emotional development. In 2019, Texas had nearly twice the average rate of loss to follow-up (LFU) or loss to documentation (LTD; i.e., missing documentation of services received) among infants who did not pass their newborn hearing screening compared to the United States overall (51.1 vs. 27.5%). We aimed to identify factors contributing to LFU/LTD among infants who do not pass their newborn hearing screening in Texas. Study Design Data were collected through semistructured qualitative interviews with 56 providers along the hearing care continuum, including hospital newborn hearing screening program staff, audiologists, primary care physicians, and early intervention (EI) program staff located in three rural and urban public health regions in Texas. Following recording and transcription of the interviews, we used qualitative data analysis software to analyze themes using a conventional content analysis approach. Results Frequently cited barriers included problems with family access to care, difficulty contacting patients, problems with communication between providers and referrals, lack of knowledge among providers and parents, and problems using the online reporting system. Providers in rural areas more often mentioned problems with family access to care and contacting families compared to providers in urban areas. Conclusion These findings provide insight into strategies that public health professionals and health care providers can use to work together to help further increase the number of children identified early who may benefit from EI services. Key Points

Obstetrics & Gynecology↗

DOE OSTI2020

A pilot study: Auditory steady-state responses (ASSR) can be measured in human fetuses using fetal magnetoencephalography (fMEG)

Auditory steady-state responses (ASSRs) are ongoing evoked brain responses to continuous auditory stimuli that play a role for auditory processing of complex sounds and speech perception. Transient auditory event-related responses (AERRs) have previously been recorded using fetal magnetoencephalography (fMEG) but involve different neurological pathways. Previous studies in children and adults demonstrated that the cortical components of the ASSR are significantly affected by state of consciousness and by maturational changes in neonates and young infants. To our knowledge, this is the first study to investigate ASSRs in human fetuses. 47 fMEG sessions were conducted with 24 healthy pregnant women in three gestational age groups (30–32 weeks, 33–35 weeks and 36–39 weeks). The stimulation consisted of amplitude-modulated (AM) tones with a duration of one second, a carrier frequency (CF) of 500 Hz and a modulation frequency (MF) of 27 Hz or 42 Hz. Both tones were presented in a random order with equal probability adding up to 80–100 repetitions per tone. The ASSR across trials was quantified by assessing phase synchrony in the cortical signals at the stimulation frequency. Ten out of 47 recordings were excluded due to technical problems or maternal movements. Analysis of the included 37 fetal recordings revealed a statistically significant response for the phase coherence between trials for the MF of 27 Hz but not for 42 Hz. An exploratory subgroup analysis moreover suggested an advantage in detectability for fetal behavioral state 2F (active asleep) compared to 1F (quiet asleep) detected using fetal heart rate. In conclusion, this pilot study is the first description of a method to detect human ASSRs in fetuses. The findings warrant further investigations of the developing fetal brain.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2023

Prevalence of Self-Reported Bothersome Tinnitus in U.S. Army Soldiers From January 1, 2015, Through September 30, 2019

Purpose: Tinnitus is a symptom of auditory dysfunction or injury and can be a precursor to permanent hearing loss. Tinnitus can interfere with communication, sleep, concentration, and mood; when this occurs, it is referred to as bothersome tinnitus. Annual hearing surveillance in the U.S. Army includes screening for bothersome tinnitus. Estimating the prevalence of self-reported bothersome tinnitus can facilitate prioritization of prevention and education efforts. The objective of this study was to examine Army hearing conservation data to estimate the prevalence of self-reported bothersome tinnitus as a function of age, hearing, sex, service component, and pay grade. Method: The study employed a cross-sectional, retrospective design. Records from 1,485,059 U.S. Army Soldiers retrieved from the Defense Occupational and Environmental Health Readiness System–Hearing Conservation were analyzed. Descriptive statistics and multinomial logistic regression analysis were used to estimate the prevalence of bothersome tinnitus and its associations with Soldiers' demographic characteristics. Results: The estimated prevalence of self-reported bothersome tinnitus was 17.1% between January 1, 2015, and September 30, 2019; 13.6% of the Soldiers reported “bothered a little” and 3.5% reported “bothered a lot.” Proportionally, the prevalence of self-reported bothersome tinnitus was higher for males, older Soldiers, and Reserve Component Soldiers. For every 1-year increase in age, the odds of self-reporting “bothered a little” tinnitus relative to “bothered not at all tinnitus” would be expected to increase by 2.2% (2.1%, 2.3%), and the odds of self-reporting “bothered a lot” tinnitus relative to “bothered not at all” tinnitus would be expected to increase by 3.6% (3.5%, 3.7%). Conclusions: The self-reported prevalence of bothersome tinnitus in the U.S. Army (17.1%) is substantially higher than that of the general population, where prevalence of bothersome tinnitus is estimated at 6.6%. Examination of bothersome tinnitus among Soldiers is an important step toward optimizing prevention, education, and intervention efforts.

Audiology & Speech-Language Pathology↗

DOE OSTI2024

Normative Ranges for Oculomotor and Reaction Time Tests in U.S. Military Service Members and Veterans

AbstractBackground Oculomotor and reaction time tests are frequently used assessments of vestibular symptoms, traumatic brain injury (TBI), or other neurological disorders in both clinical and research contexts. When interpreting these tests it is important to have a reference interval (RI) as a comparison for what constitutes a typical/expected response; however, the current body of research has only limited information regarding normative ranges calculated according to established standards or for a military-specific sample.Purpose The purpose of the present study was to describe RIs for oculomotor and reaction time tests in a cohort of service members and veterans (SMVs) for use as comparators by clinicians and scientists.Research Design Descriptive.Study Sample Participants were prospectively enrolled in the Defense and Veterans Brain Injury Center-Traumatic Brain Injury Center of Excellence 15-year Longitudinal Traumatic Brain Injury Study. Only SMVs without a history of TBI or blast exposure were included in the RI calculations.Data Collection and Analysis The test paradigms included in this analysis were: smooth pursuit, prosaccades, antisaccades, saccades and reaction time, predictive saccades, optokinetic nystagmus, auditory reaction time, and visual reaction time. Nonparametric methods, based on the U.S. Food and Drug Administration's recognized consensus standards, were used to calculate 95% RIs. A comparison between the calculated RIs and those available from previously published research is provided.Results Summary statistics and RIs were calculated for 47 outcome parameters from 13 oculomotor and reaction time tests. Sample sizes and age ranges varied across outcome parameters depending on the availability of reference values for RI calculations. The sample sizes used to calculate RIs ranged from 51 to 69. The age of SMVs included in each RI ranged from 19 to 61 years with mean ages ranging from 37 to 39 years. Similarities/differences between the RIs in the present study and those in previously published research are highly dependent on the outcome parameter; however, in general, the RIs in the present study tended to be somewhat wider.Conclusion The RIs provided in this paper can serve as comparisons for clinicians and scientists who are utilizing these oculomotor and reaction time testing paradigms in similar cohorts of patients or research participants.

Audiology & Speech-Language Pathology↗