Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “SPEECH”

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.

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

At least 91 records · Page 5

Automatic speech recognition technology development at ITT Defense Communications Division

An assessment of the applications of automatic speech recognition to defense communication systems is presented. Future research efforts include investigations into the following areas: (1) dynamic programming; (2) recognition of speech degraded by noise; (3) speaker independent recognition; (4) large vocabulary recognition; (5) word spotting and continuous speech recognition; and (6) isolated word recognition.

White, George M.↗

Index to NASA news releases and speeches, 1993

This issue of the Index to NASA News Releases and Speeches contains a listing of news releases distributed by the Office of Public Affairs, NASA Headquarters, and a selected listing of speeches presented by members of the Headquarters staff during 1993. The index is arranged in six sections: subject index, personal names index, news release number index, accession number index, speeches, and news releases.

Source record↗

Watch what you say, your computer might be listening: A review of automated speech recognition

Spoken language is the most convenient and natural means by which people interact with each other and is, therefore, a promising candidate for human-machine interactions. Speech also offers an additional channel for hands-busy applications, complementing the use of motor output channels for control. Current speech recognition systems vary considerably across a number of important characteristics, including vocabulary size, speaking mode, training requirements for new speakers, robustness to acoustic environments, and accuracy. Algorithmically, these systems range from rule-based techniques through more probabilistic or self-learning approaches such as hidden Markov modeling and neural networks. This tutorial begins with a brief summary of the relevant features of current speech recognition systems and the strengths and weaknesses of the various algorithmic approaches.

Degennaro, Stephen V.↗

Vector adaptive predictive coder for speech and audio

A real-time vector adaptive predictive coder which approximates each vector of K speech samples by using each of M fixed vectors in a first codebook to excite a time-varying synthesis filter and picking the vector that minimizes distortion. Predictive analysis for each frame determines parameters used for computing from vectors in the first codebook zero-state response vectors that are stored at the same address (index) in a second codebook. Encoding of input speech vectors s.sub.n is then carried out using the second codebook. When the vector that minimizes distortion is found, its index is transmitted to a decoder which has a codebook identical to the first codebook of the decoder. There the index is used to read out a vector that is used to synthesize an output speech vector s.sub.n. The parameters used in the encoder are quantized, for example by using a table, and the indices are transmitted to the decoder where they are decoded to specify transfer characteristics of filters used in producing the vector s.sub.n from the receiver codebook vector selected by the vector index transmitted.

Chen, Juin-Hwey↗

Applications for Subvocal Speech

A research and development effort now underway is directed toward the use of subvocal speech for communication in settings in which (1) acoustic noise could interfere excessively with ordinary vocal communication and/or (2) acoustic silence or secrecy of communication is required. By "subvocal speech" is meant sub-audible electromyographic (EMG) signals, associated with speech, that are acquired from the surface of the larynx and lingual areas of the throat. Topics addressed in this effort include recognition of the sub-vocal EMG signals that represent specific original words or phrases; transformation (including encoding and/or enciphering) of the signals into forms that are less vulnerable to distortion, degradation, and/or interception; and reconstruction of the original words or phrases at the receiving end of a communication link. Potential applications include ordinary verbal communications among hazardous- material-cleanup workers in protective suits, workers in noisy environments, divers, and firefighters, and secret communications among law-enforcement officers and military personnel in combat and other confrontational situations.

Jorgensen, Charles↗

Effect of Whole-Body Vibration on Speech: Effect on Intelligibility - Part 2

The effect on speech intelligibility was measured for speech where talkers reading Diagnostic Rhyme Test material were exposed to 0.7 g whole body vibration to simulate space vehicle launch. Across all talkers, the effect of vibration was to degrade the percentage of correctly transcribed words from 83% to 74%. The magnitude of the effect of vibration on speech communication varies between individuals, for both talkers and listeners. A worst case scenario for intelligibility would be the most sensitive listener hearing the most sensitive talker; one participant s intelligibility was reduced by 26% (97% to 71%) for one of the talkers.

Begault, Durand R.↗

Pilot Workload and Speech Analysis: A Preliminary Investigation

Prior research has questioned the effectiveness of speech analysis to measure the stress, workload, truthfulness, or emotional state of a talker. The question remains regarding the utility of speech analysis for restricted vocabularies such as those used in aviation communications. A part-task experiment was conducted in which participants performed Air Traffic Control read-backs in different workload environments. Participant's subjective workload and the speech qualities of fundamental frequency (F0) and articulation rate were evaluated. A significant increase in subjective workload rating was found for high workload segments. F0 was found to be significantly higher during high workload while articulation rates were found to be significantly slower. No correlation was found to exist between subjective workload and F0 or articulation rate.

Bittner, Rachel M.↗

Speech Alarms Pilot Study

Speech alarms have been used extensively in aviation and included in International Building Codes (IBC) and National Fire Protection Association's (NFPA) Life Safety Code. However, they have not been implemented on space vehicles. Previous studies conducted at NASA JSC showed that speech alarms lead to faster identification and higher accuracy. This research evaluated updated speech and tone alerts in a laboratory environment and in the Human Exploration Research Analog (HERA) in a realistic setup.

Sandor, Aniko↗

Trusted Communication: Utilizing Speech Communication to Enhance Human-Machine Teaming Success

An area of increasing interest for the next generation of aircraft is autonomy and the integration of increasingly autonomous systems into the national airspace. Such an integration requires humans to work closely with autonomous systems, forming teams. Our hypothesis is that a team composed of both humans and autonomous systems will operate better than either entity alone. We have existing procedures for certifying pilots to operate in the national airspace and are currently working on methods for validating the function of autonomous systems, however we have no method in place for assessing the interaction of these two disparate systems. Communication is one avenue. This paper will examine the use of language as a metric for ascertaining human-machine teaming effectiveness. A proof-of-concept of the application of two communication-based analysis techniques, Linguistic Inquiry and Word Count (LIWC) and Latent Semantic Analysis (LSA), for the prediction of success in human/chatbot teaming was conducted. By running these analyses over data from the 2014 and 2015 Loebner Prize competitions of human/chatbot teaming, numerical scores were obtained that can be associated with scores provided by human judges during the competition. Correlating their LIWC and LSA data with the scores provided by the judges, and using linear regression over this correlation, formulae were obtained that predict the score of human/chatbot interaction. These formulae were tested over the 2013 Loebner Prize transcripts, determining that, though there was strong correlation between predicted and actual scores, the predictive success of this method was not strong. However, with specialized topic spaces and lexica, as well as larger data sets, the predictive power of these metrics will improve. Given the importance of providing metrics for human-machine system team success and given the promise shown by the communication-basedLIWCand LSAmethods, continuing research in this area is necessary. After examining the potential for using communication and spoken language as a metric for the success of human/autonomous system teaming, this paper then examines aspects inherent to communication systems that may contribute to unreliability and reduced trust. Modern natural language processing tools rely on deep learning algorithms to create language rules that produce accurate results, but these rules are uninterpretable. The resulting blackbox system lacks transparency necessary for full validation and complete trust. Additionally, speech-based interfaces pose other difficulties to developing coordinated teamwork between humans and autonomous systems. Human communication is infrequently limited to speech only, instead usually relying on a combination of verbal, gestural, and general body language communication. Reducing an analysis of team effectiveness to a study of spoken language alone is problematic as it leaves these other equally important forms of communication out. This paper will examine these problems and the general deficiencies in speech-based metrics for human-machine teaming.

E L Meszaros↗

Automatic Speech Recognition for Launch Control Center Communication Using Recurrent Neural Networks with Data Augmentation and Custom Language Model

Transcribing voice communications in NASA’s launch control center is important for information utilization. However, automatic speech recognition in this environment is particularly challenging due to the lack of training data, unfamiliar words in acronyms, multiple different speakers and accents, and conversational characteristics of speaking. We used bidirectional deep recurrent neural networks to train and test speech recognition performance. We showed that data augmentation and custom language models can improve speech recognition accuracy. Transcribing communications from the launch control center will help the machine analyze information and accelerate knowledge generation.

Chow, Edward↗

Speech Emotion Feature Selection Method Based on Contribution Analysis Algorithm of Neural Network

There are many emotion features. If all these features are employed to recognize emotions, redundant features may be existed. Furthermore, recognition result is unsatisfying and the cost of feature extraction is high. In this paper, a method to select speech emotion features based on contribution analysis algorithm of NN is presented. The emotion features are selected by using contribution analysis algorithm of NN from the 95 extracted features. Cluster analysis is applied to analyze the effectiveness for the features selected, and the time of feature extraction is evaluated. Finally, 24 emotion features selected are used to recognize six speech emotions. The experiments show that this method can improve the recognition rate and the time of feature extraction.

97 MATHEMATICS AND COMPUTING↗

Speech communications

Analysis and synthesis of speech, and use of procedures for studying human speech processes

SPEECH↗

Speech communication

Linguistic and speech communication concepts on mechanism of glottal vibration for vowels and consonants, and homomorphic speech processing

COMMUNICATION SYSTEM↗

Multichannel spatial auditory display for speech communications

A spatial auditory display for multiple speech communications was developed at NASA/Ames Research Center. Input is spatialized by the use of simplified head-related transfer functions, adapted for FIR filtering on Motorola 56001 digital signal processors. Hardware and firmware design implementations are overviewed for the initial prototype developed for NASA-Kennedy Space Center. An adaptive staircase method was used to determine intelligibility levels of four-letter call signs used by launch personnel at NASA against diotic speech babble. Spatial positions at 30 degrees azimuth increments were evaluated. The results from eight subjects showed a maximum intelligibility improvement of about 6-7 dB when the signal was spatialized to 60 or 90 degrees azimuth positions.

NASA Discipline Space Human Factors↗

Retrospective Analysis of Clinical Performance of an Estonian Speech Recognition System for Radiology: Effects of Different Acoustic and Language Models

The aim of this study was to analyze retrospectively the influence of different acoustic and language models in order to determine the most important effects to the clinical performance of an Estonian language-based non-commercial radiology-oriented automatic speech recognition (ASR) system. An ASR system was developed for Estonian language in radiology domain by utilizing open-source software components (Kaldi toolkit, Thrax). The ASR system was trained with the real radiology text reports and dictations collected during development phases. The final version of the ASR system was tested by 11 radiologists who dictated 219 reports in total, in spontaneous manner in a real clinical environment. The audio files collected in the final phase were used to measure the performance of different versions of the ASR system retrospectively. ASR system versions were evaluated by word error rate (WER) for each speaker and modality and by WER difference for the first and the last version of the ASR system. Total average WER for the final version throughout all material was improved from 18.4% of the first version (v1) to 5.8% of the last (v8) version which corresponds to relative improvement of 68.5%. WER improvement was strongly related to modality and radiologist. In summary, the performance of the final ASR system version was close to optimal, delivering similar results to all modalities and being independent on user, the complexity of the radiology reports, user experience, and speech characteristics.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Automatic speech recognition predicts contemporaneous earthquake fault displacement

Abstract Significant progress has been made in probing the state of an earthquake fault by applying machine learning to continuous seismic waveforms. The breakthroughs were originally obtained from laboratory shear experiments and numerical simulations of fault shear, then successfully extended to slow-slipping faults. Here we apply the Wav2Vec-2.0 self-supervised framework for automatic speech recognition to continuous seismic signals emanating from a sequence of moderate magnitude earthquakes during the 2018 caldera collapse at the Kīlauea volcano on the island of Hawai’i. We pre-train the Wav2Vec-2.0 model using caldera seismic waveforms and augment the model architecture to predict contemporaneous surface displacement during the caldera collapse sequence, a proxy for fault displacement. We find the model displacement predictions to be excellent. The model is adapted for near-future prediction information and found hints of prediction capability, but the results are not robust. The results demonstrate that earthquake faults emit seismic signatures in a similar manner to laboratory and numerical simulation faults, and artificial intelligence models developed for encoding audio of speech may have important applications in studying active fault zones.

58 GEOSCIENCES↗

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 ↗