Real-time Voice Research Articles

Research and Implementation of Real-Time Voice Control System based on Android

At present most of the mobile phone users must have to constantly change their mobile phone’s contextual model to adapt to different occasions in every busy life. In order to solve this problem. This paper designed and implemented the real time auto control system of mobile phone’ voice based on Android platform. The paper introduces the sound overall framework of real-time control system around three functional modules: the sound acquisition, getting sound’s loudness and adjusting ring and expounded key technology of the system in the implementation process. The result of experiment shows that the system can change ringtone loudness on the degree of noisy environment in real time and the system run reliably, meanwhile it has certain applicability and practical value.

Arabic Command Based Human Computer Interaction

This paper means to build up a strategy for helping debilitated individuals old and amputee, by using unimodal human computer interaction; real time voice command and interaction between patient and computer where these mixes give a promising answer for help the incapacitated individuals. The fundamental goal of the work is to design a project that deals easily with a patient that uses Arabic commands. The project is based on training a dataset that collected from different persons that have ages from 12-58 years old, unknown word and background noise. The propose system can works with different direction in Arabi language (“اذهب“, ” يمين“, ”يسار“, ”قف”) and with a multispeed that based on pulse width modulation (PWM). These functions based on code written in MATLAB environment. The project uses ATmega328microcontroller.

Learning Efficient Representations for Fake Speech Detection

Synthetic speech or “fake speech” which matches personal vocal traits has become better and cheaper due to advances in deep learning-based speech synthesis and voice conversion approaches. This increased accessibility of synthetic speech systems and the growing misuse of them highlights the critical need to build countermeasures. Furthermore, new synthesis models evolve all the time and the efficacy of previously trained detection models on these unseen attack vectors is poor. In this paper, we focus on: 1) How can we build highly accurate, yet parameter and sample-efficient models for fake speech detection? 2) How can we rapidly adapt detection models to new sources of fake speech? We present four parameter-efficient convolutional architectures for fake speech detection with best detection F1 scores of around 97 points on a large dataset of fake and bonafide speech. We show how the fake speech detection task naturally lends itself to a novel multi-task problem further improving F1 scores for a mere 0.5% increase in model parameters. Our multi-task setting also helps in data-sparse situations, commonplace in adversarial settings. We investigate an alternative approach to the data-sparsity problem using transfer learning and show that it is possible to meet purely supervised detection performance for unseen attack vectors with as little as 6.25% of the training data. This is the first known application of transfer learning in adversarial settings for speech. Finally, we show how well our transfer learning approach adapts in an instance-efficient way to new attack vectors using the Real-Time Voice Cloning toolkit. We exceed the purely supervised detection performance (99.18 F1) with as little as 6.25% of the data.

Implementation of Class-Based Low Latency Fair Queueing (CBLLFQ) Packet Scheduling Algorithm for HSDPA Core Network

To provide a guaranteed Quality of Service (QoS) to real-time traffic in High-Speed Downlink Packet Access (HSDPA) core network, we proposed an enhanced mechanism. For an enhanced QoS, a Class-Based Low Latency Fair Queueing (CBLLFQ) packet scheduling algorithm is introduced in this work. Packet classification, metering, queuing, and scheduling using differentiated services (DiffServ) environment was the points in focus. To classify different types of real-time voice and multimedia traffic, the QoS provisioning mechanisms use different DiffServ code points (DSCP).The proposed algorithm is based on traffic classes which efficiently require the guarantee of services and specified level of fairness. In CBLLFQ, a mapping criterion and an efficient queuing mechanism for voice, video and other traffic in separate queues are used. It is proved, that the algorithm enhances the throughput and fairness along with a reduction in the delay and packet loss factors for smooth and worst traffic conditions. The results calculated through simulation show that the proposed calculations meet the QoS prerequisites efficiently.

Neural Correlates of Vocal Pitch Compensation in Individuals Who Stutter.

Stuttering is a disorder that impacts the smooth flow of speech production and is associated with a deficit in sensorimotor integration. In a previous experiment, individuals who stutter were able to vocally compensate for pitch shifts in their auditory feedback, but they exhibited more variability in the timing of their corrective responses. In the current study, we focused on the neural correlates of the task using functional MRI. Participants produced a vowel sound in the scanner while hearing their own voice in real time through headphones. On some trials, the audio was shifted up or down in pitch, eliciting a corrective vocal response. Contrasting pitch-shifted vs. unshifted trials revealed bilateral superior temporal activation over all the participants. However, the groups differed in the activation of middle temporal gyrus and superior frontal gyrus [Brodmann area 10 (BA 10)], with individuals who stutter displaying deactivation while controls displayed activation. In addition to the standard univariate general linear modeling approach, we employed a data-driven technique (independent component analysis, or ICA) to separate task activity into functional networks. Among the networks most correlated with the experimental time course, there was a combined auditory-motor network in controls, but the two networks remained separable for individuals who stuttered. The decoupling of these networks may account for temporal variability in pitch compensation reported in our previous work, and supports the idea that neural network coherence is disturbed in the stuttering brain.

Newly Designed Audio-Visual Environment for Laryngeal Examination with Real-Time Voice and Acoustic Analysis Displayed on Multiple Monitors under Various Platforms

喉頭画像・音声・声の基本周波数・音響解析結果等を時間軸を一致させて記録保存し結果を二次利用することが，誰でも簡単に行えるようになった。検査をする側の技術的な問題は解決されたため，次に，検査の説明を受ける側（被検者，家族，医療関係者，医師等）の環境整備が必要と考えられた。喉頭動画像および解析結果（声の基本周波数，音響解析結果等）を，各個人が自由に同時にリアルタイムに見られるようにしておく必要がある。今回われわれは，インターネット関連技術のWebRTCとH.264等を用いることにより，一般的に使用されているWi-Fiを用いて，iOS, Androidいずれの携帯端末が混在している環境でもリアルタイムに動画像が共有できることを確認した。

SPQER: Speech Quality Evaluation Using Word Recognition for VoIP Communication in Lossy and Mobile Networks

In this paper, we introduce SPQER (pronounced speaker), a novel approach to evaluate the quality of experience for real-time Voice over IP (VoIP) communication in mobile and lossy networks. Traditional speech quality metrics, e.g., Perceptual Evaluation of Speech Quality (PESQ) or the Hearing-Aid Speech Quality Index (HASQI), directly compare frequencies and amplitudes to calculate the received signal distortions. SPQER instead uses machine learning classification to evaluate the percentage of recognizable words in conjunction with a time-based decay function to penalize delay and cross-talking. So instead of evaluating noise, SPQER directly answers the question: What percentage of words is the recipient able to understand? We presented a sensitivity analysis, which is based on testbed experiments for different packet loss rates and simulated delays, to asses the impact of challenging link conditions. A final correlation analysis to a short user study shows that SPQER can better evaluate the amount of understandable words than PESQ and HASQI, while still giving a more precise indication about the voice quality than the Word Error Rate (WER) metric.

Real-time voice processing system for the loudspeakers under noisy environments

Real-time voice processing system for the loudspeakers under noisy environments

Real-Time Voice Transmission over Wireless Sensor Network (VoWSN) based Automatic Speech Recognition (ASR) Technique

Real-Time Voice Transmission over Wireless Sensor Network (VoWSN) based Automatic Speech Recognition (ASR) Technique

Moderation Challenges in Voice-based Online Communities on Discord

Online community moderators are on the front lines of combating problems like hate speech and harassment, but new modes of interaction can introduce unexpected challenges. In this paper, we consider moderation practices and challenges in the context of real-time, voice-based communication through 25 in-depth interviews with moderators on Discord. Our findings suggest that the affordances of voice-based online communities change what it means to moderate content and interactions. Not only are there new ways to break rules that moderators of text-based communities find unfamiliar, such as disruptive noise and voice raiding, but acquiring evidence of rule-breaking behaviors is also more difficult due to the ephemerality of real-time voice. While moderators have developed new moderation strategies, these strategies are limited and often based on hearsay and first impressions, resulting in problems ranging from unsuccessful moderation to false accusations. Based on these findings, we discuss how voice communication complicates current understandings and assumptions about moderation, and outline ways that platform designers and administrators can design technology to facilitate moderation.

Dialogue System of Controlling Robot Based on the Theory of Finite-State Automata

The article discusses the system of dialogue control manipulation robots. The analysis of the basic methods of automatic speech recognition, speech understanding, dialogue management, voice response synthesis in dialogue systems has been carried out. Three types of dialogue management are considered as "system initiative", "user initiative" and "combined initiative". A system of object-oriented dialog control of a robot based on the theory of finite state machines with using a deep neural network is proposed. The main difference of the proposed system lies in the separate implementation of the dialogue process and robot’s actions, which is close to the pace of natural dialogue control. This method of constructing a dialogue control robot allows system to automatically correct the result of speech recognition, robot’s actions based on tasks. The necessity of correcting the result of speech recognition and robot’s actions may be caused by the users’ accent, working environment noise or incorrect voice commands. The process of correcting speech recognition results and robot’s actions consists of three stages, respectively, in a special mode and a general mode. The special mode allows users to directly control the manipulator by voice commands. The general mode extends the capabilities of users, allowing them to get additional information in real time. At the first stage, continuous speech recognition is built by using a deep neural network, taking into account the accents and speech speeds of various users. Continuous speech recognition is a real-time voice to text conversion. At the second stage, the correction of the speech recognition result by managing the dialogue based on the theory of finite automata. At the third stage, the actions of the robot are corrected depending on the operating state of the robot and the dialogue management process. In order to realize a natural dialogue between users and robots, the problem is solved in creating a small database of possible dialogues and using various training data. In the experiments, the dialogue system is used to control the KUKA manipulator (KRC4 control) to put the desired block in the specified location, implemented in the Python environment using the RoboDK software. The processes and results of experiments confirming the operability of the interactive robot control system are given. A fairly high accuracy (92 %) and an automatic speech recognition rate close to the rate of natural speech were obtained.

Campus bullying detection based on motion recognition and speech emotion recognition

In many areas, school bullying incidents often occur. Such incidents not only damage the learning environment of the school but also cause great harm to the students who are bullied. The occurrence of the campus bullying incident is generally hidden and not easy to be discovered by teachers or parents in time. Aiming at the problem that the campus bullying incident needs to be discovered in time, a campus bullying detection method based on motion recognition and speech emotion recognition is proposed. The electronic device such as smart phone and smart watch worn by people are used to collect the data of human motion and voice in real time. Thus, it is possible to detect in time whether the wearer is being bullied. In this paper, six characteristics of human motion data and MFCC features of speech data are extracted. Then the PCA algorithm is used to reduce the dimension of the obtained feature matrix. Finally, the KNN algorithm is used to form the motion and speech recognition network. After cross-validation, the average recognition rate of the bullying movement and speech emotion of the system in this paper are 77.8% and 81.4%, respectively. The experimental results show that the campus bullying detection method based on KNN algorithm has obtained a good recognition rate.

Towards artificial general intelligence with hybrid Tianjic chip architecture.

There are two general approaches to developing artificial general intelligence (AGI)1: computer-science-oriented and neuroscience-oriented. Because of the fundamental differences in their formulations and coding schemes, these two approaches rely on distinct and incompatible platforms2-8, retarding the development of AGI. A general platform that could support the prevailing computer-science-based artificial neural networks as well as neuroscience-inspired models and algorithms is highly desirable. Here we present the Tianjic chip, which integrates the two approaches to provide a hybrid, synergistic platform. The Tianjic chip adopts a many-core architecture, reconfigurable building blocks and a streamlined dataflow with hybrid coding schemes, and can not only accommodate computer-science-based machine-learning algorithms, but also easily implement brain-inspired circuits and several coding schemes. Using just one chip, we demonstrate the simultaneous processing of versatile algorithms and models in an unmanned bicycle system, realizing real-time object detection, tracking, voice control, obstacle avoidance and balance control. Our study is expected to stimulate AGI development by paving the way to more generalized hardware platforms.

Safe2Walk4Blind: 시각장애인의 안전 보행을 위한 심층 신경망 기반의 보행 보조 시스템

This paper describes the hardware and software co-design of deep-learning-based smart glasses that enable blind people to recognize dangerous obstacles in real time to allow them to walk safely and independently. We use a DNN(Deep Neural Network) model to detect objects and estimate distances, which autonomously alerts blind pedestrians to emergency situations. The proposed Safe2Walk4Blind(Safe to Walk for the Blind) system adopts a server-client architecture to overcome the restricted computing resource of the embedded glasses. The system processes the streaming video, captured by the smart glasses with a monocular camera, and sends the classification and distance estimates of the encountered obstacle to the blind by voice in real time. Our experiments show that the Safe2Walk4Blind system can recognize objects with an accuracy of 82.2% and estimate distances with an accuracy of 89.6% while processing data at 30 frames per second.

Survey on Detection Techniques for Denial-of-Service Attacks in SIP-Based Voice Over IP Networks

VoIP provides real-time voice communications over packet-switched networks. VoIP is widely used as it offers more flexibility when compared to the traditional Public Switched Telephone Network (PSTN). Session Initiation Protocol (SIP) is a protocol used for initiating a session between the caller and the callee in the VoIP system. SIP is a text-oriented protocol and is implemented in a susceptible environment. These features make SIP protocol vulnerable to various security risks such as Denial of Service, Registration hijacking and Eavesdropping. Detection of these attacks plays a vital role. The Survey focuses on different techniques used for detection of DoS attacks and differentiating between the proposed mechanisms for identifying abnormalities and then notifying the end-users about the possible intrusions.

Difference-Set-Based Channel Hopping for Minimum-Delay Blind Rendezvous in Multi-Radio Cognitive Radio Networks

Rendezvous in the blind fashion is known as one of the most robust and effective ways for cognitive users (CUs) to exchange control information under the dynamic interference of primary users (PUs). Earlier, no existing scheme could guarantee CUs to rendezvous blindly in each timeslot, which makes them experience relatively long exchange delay. Based on the mathematical concept of relaxed cyclic difference set ( DS ), this paper proposes two novel channel hopping (CH) schemes for multi-radio CUs with and without a synchronized time clock to rendezvous blindly in every timeslot. This achievement of minimum rendezvous delay is especially useful for such delay-sensitive applications as real-time voice and emergency communications. Moreover, based on a special class of DS, the proposed two schemes can closely approach the theoretical lower bound on the number of radio transceivers for CUs to rendezvous blindly in every timeslot and, hence, minimizes their required hardware complexity to exchange control information at the minimum rendezvous delay. Numerical simulation shows that, under dynamic PU interference, the proposed schemes outperform the existing ones in terms of the average throughput and delay of CUs.

Design of Students’ Spoken English Pronunciation Training System Based on Computer VB Platform

Spoken English communication is most commonly used in the international communication. However, the accuracy of spoken English pronunciation is the key factor to restrict English learners in China. For the current situation that spoken English proficiency is generally low in China, this paper aims to design a spoken English pronunciation training system that will provide guidance and help for English learners’ spoken pronunciation. The Visual Basic platform is used in the design of the system. This paper first conducts an in-depth study on the related theories of voice recognition, discusses the correction algorithm of voice scoring and pronunciation, and puts forward more practical and convenient AP-based scoring method, providing full theoretical support for the design of the system. Then through the function analysis and design of the spoken English pronunciation training system, this paper realizes the system design of scoring and correcting errors of English spoken pronunciation based on the VB platform. The system boasts the basic functions, including English phonetic symbols and word pronunciation to follow, real-time voice evaluation, and pronunciation error correction. According to the test, the similarity of the system with the experts is over 90% in scoring and its efficiency of pronunciation error correction reaches 80%, which plays a certain role in improving spoken English of English learners.

Embedded Parallelism Enabling Ultralow-Power Zigbee Voice Communications

Short-range wireless technologies are known to transmit voice, audio, image, and video messages in real time. Energy consumption and transmission reach are critical in such networks, especially for portable and power autonomous devices. The purpose of the Voice over Zigbee technology is to provide a competitive offering that excels in these performance aspects. Due to the CSMA-CA mechanism implemented in the 802.15.4 layer, a well-designed strategy must be considered in Zigbee to create a robust, reliable, and full-duplex conversation. In past efforts, we proved that the radio channel of Zigbee has enough bandwidth to support a full-duplex conversation with narrow-band voice codecs. Our embedded implementation of the Speex voice codec targeted the development of a low-cost, ultralow-power, long-range wireless headset using Zigbee technology to transmit voice in full-duplex mode for use with leading PC VoIP programs. Furthermore, we presented the real environment performance evaluation and power consumption tests involving the developed headset prototype. Talk time is comparable to Bluetooth including at the power budget, the codec processing, and analogue audio interface, but its deep-sleep lifetime more than doubles the Bluetooth performance. This was one of the very few successful efforts to port a voice codec on an ultralow-power DSP for use with power sensitive Zigbee applications, which is highly cited in the literature, proving additionally that using an open-source codec can deliver similar voice quality, reducing the total system cost. The current paper elaborates on the embedded parallelism of the Speex implementation and the exploitation of the DSP architectural parallelism which critically enabled the Voice over Zigbee application on the ultralow-power DSP platform. Another significant contribution of this work is towards understanding and resolving the challenges faced when trying to achieve good quality transmission of media over constrained devices and networks. The way to new ultralow-power voice-related Zigbee and constrained network applications is open.

Design and Development of Secure Mobile Communication over GSM Network Using Open Source Operating System (OS)

With the rapidly advancing technology of today, exchange of information and data is a very pertinent matter. The world has just recently witnessed the effects of information leakage through the issue of WikiLeaks. There are huge amounts of data being shared over different platforms nowadays. Global System for Mobile Communication (GSM) is one of the most reliable platforms known to and used by almost all people in the world for text as well as voice communication. With the tools like Android Studio and NetBeans available, it is now possible to encrypt the text that has to be sent over the GSM, so that it can be decrypted at the other end of the communication path. However, the encryption and decryption of voice being transmitted over the GSM network still remains a question. In the domain of real time voice encryption, much of the work being carried out pertains to the voice being exchanged through the Internet Protocol. As compared to the Voice over Internet Protocol (VoIP), voice over the GSM network has not seen much research work related to its security aspects. The purpose of this paper is to document the results of a project aimed at developing a platform for mobile phones in order to communicate over the GSM network in a secure manner. The most suitable method for achieving the above mentioned objective is to use an open source Operating System (OS), so that the source code is easily accessible and usable. In this paper, the Android OS will be under discussion, which is compatible with all the Android mobile phones. In this way, the maximum number of mobile phone users can be benefitted because Android cell phones are being widely used nowadays. The use of cryptographic algorithms for securing the voice communication over the GSM network is also a part of this paper. The work revolves around the Java programming language since the Android application development has been carried out in Java through the use of Android Studio. Also, NetBeans has been employed for developing algorithms for voice encryption

Effect of Different Feedback Modalities on Swimming Pace: Which Feedback Modality is Most Effective?

To compare the effect of three different feedback modalities on swimming pace, sixteen male swimmers and triathletes participated in this study. Each participant swam 3 x 400 m, one for each feedback modality, swimming front crawl at 80% of their individual swimming critical speed. Three feedback modalities were examined: self-pacing, real time visual feedback and real time voice feedback. The swimmers adopted a fast start in all feedback modalities. In the real time voice feedback modality, the data recorded during the second lap (200 m) showed a significant improvement of their swimming pace approaching the swimming pace intended (-1.47 s, p < .01, medium effect size 0.79). A significant improvement toward the swimming pace intended was also noticed at the third split time (300 m) (0.05 s, p < .01, large effect size 0.81) and at the fourth split time (400 m) (0.46 s, p < .01, medium effect size 0.76). In self-pacing, the swimmers were not able to swim in line with the swimming pace intended. In real time visual feedback modality, the swimmers did not show a significant improvement approaching the swimming pace intended. The results revealed that communication with the swimmers using the real time voice feedback induced a significant improvement in their swimming pace and could help the athletes to swim with accurate and consistent pace.