Search

MPEG surround is one of the most popular multichannel audio coding standards used today. In producing quality sound, there are several parts that are influential, but those that are directly related to sound reproduction, one of which is the method of downmixing, mixing the right channels will improve the quality of the sound produced. Because this research is closely related to data computing, the Matrix Laboratory (MATLAB) application is used. There are 2 stages carried out in the MATLAB application, namely the encoding and decoding process. In the encoding process in MATLAB, each audio channel is set according to a previously designed scheme. The results of this encoding are then decoded, so that the ODG value of each tested audio channel is obtained. In his presentation, the design and analysis of several domnmixing schemes was carried out. There are 5 schemes tested and to determine the performance of each scheme, these schemes are tested at high bitrates, from the range of 1000 Kbps to 2000 Kbps. Based on the test, it was found that the effect of increasing the bitrate is directly proportional to the resulting audio quality. During testing Scheme-5 has a better level of stability among all downmixing schemes and the average percentage increase in Objective Difference Grade (ODG) for all audio samples is 9.6927 %., MPEG surround merupakan salah satu standard pengkodean audio multikanal yang populer digunakan saat ini. Dalam menghasilkan suara yang berkualitas, ada beberapa bagian yang berpengaruh, namun yang berkaitan langsung dengan reproduksi suara, salah satunya adalah metode dalam melakukan downmix, percampuran kanal yang tepat akan meningkatkan kualitas dari suara yang dihasilkan. Dikarenakan penelitian ini berkaitan erat dengan komputasi data, maka digunakanlah aplikasi Matrix Laboratory (MATLAB). Ada 2 tahap yang dilakukan pada aplikasi MATLAB, yaitu proses encoding dan decoding. Pada proses encoding di MATLAB, disetting setiap kanal audio sesuai dengan skema yang telah dirancang sebelumnya. Hasil dari encoding ini kemudian di decode, sehingga didapatkan nilai ODG dari setiap kanal audio yang diujikan. Pada pemaparannya dilakukan perancangan dan analisis terhadap beberapa skema domnmixing. Terdapat 5 skema yang diujikan dan untuk mengetahui performansi dari tiap skema, skema tersebut diuji pada bitrate tinggi mulai dari range 1000 Kbps sampai dengan 2000 Kbps. Berdasarkan pengujian didapati bahwa pengaruh kenaikan bitrate berbanding lurus terhadap kualitas audio yang dihasilkan. Selama pengujian Skema-5 memiliki tingkat kestabilan yang lebih baik diantara semua skema downmixing dan rata-rata persentase kenaikan Objective Difference Grade (ODG) untuk seluruh sampel audio adalah 9,6927 %.

The article provides an interpretation of the most commonly used terms and terminological components, provides practical recommendations for standartization of variance of terms, which will be useful to experts when analyzing the results and preparing expert conclusions.

Binaural system is an audio technology which relatively easy to apply these days. This was due to the growth in technology developments of the audio recording system. The application of this system is not well known in Indonesia. This was aligned with public assumption where Wayang Kulit performance are one of many unappreciated forms of arts in Indonesia. The main problem that occurred is that whether the binaural system conjoined with the Virtual Reality (VR) System can make the public more appreciate the art form of Indonesian Wayang Kulit. By using the theory of perception conjoin with the binaural technology, it can be assumed that the audience can have a new experience from the performance through the VR thus elevating their appreciation for the art itself. This research will be divided into three stages. The first stage focused on the application of the binaural system on Wayang Kulit Animation through VR system. The second stage will focus on the product assessment whether it can elevate the appreciation of the audience for the art itself. The third stage will focus on applying the binaural system on other Indonesian art form. This year, the research will begin at the first stage.

Conversational assistants in the form of stand-alone devices such as Amazon Echo and Google Home have become popular and embraced by millions of people. By serving as a natural interface to services ranging from home automation to media players, conversational assistants help people perform many tasks with ease, such as setting timers, playing music and managing to-do lists. While these systems offer useful capabilities, they are largely passive and unaware of the human behavioral context in which they are used. In this work, we explore how off-the-shelf conversational assistants can be enhanced with acoustic-based human activity recognition by leveraging the short interval after a voice command is given to the device. Since always-on audio recording can pose privacy concerns, our method is unique in that it does not require capturing and analyzing any audio other than the speech-based interactions between people and their conversational assistants. In particular, we leverage background environmental sounds present in these short duration voice-based interactions to recognize activities of daily living. We conducted a study with 14 participants in 3 different locations in their own homes. We showed that our method can recognize 19 different activities of daily living with average precision of 84.85% and average recall of 85.67% in a leave-one-participant-out performance evaluation with 30-second audio clips bound by the voice interactions.

Background: In forensic science the process of proving authenticity of audio recording plays an important role. In recent times, Forensic experts mostly receives digital recording for authentication as compared to analog recording. A digitally altered audio signal, leaves no visual indications of being tampered, and it will be indistinguishable from an original audio signal. Objective: To highlight the significance of latency feature of mobile phone handsets in forensic science via comparing input audio latency feature of Samsung and Motorola mobile phone in two audio formats. Methods: In this work two wellestablished and most used brands of mobile phones were considered for comparison: SAMSUNG and MOTOROLA. In the present paper, the digital audio samples have been recorded using 20 mobile phones of various models from two different makes i.e. SAMSUNG and MOTOROLA, in two audio formats i.e. WAV and 3GP. Audio samples were then analysed using Adobe Audition 3.0 software for the input audio latency feature of mobile phones and compared. Findings: Input audio latency value of digital audio recordings can be helpful in forensic identification of make and model of source mobile phone. Novelty: A new technique in digital forensics, to classify the given audio samples on the basis of input audio latency feature and identifying the make of source mobile handsets. Keywords: Authentication; digital audio; forensic science; adobe audition; mobile phone

With the popularity of mobile devices and the emergence of various audio-editing tools, it becomes easier to produce and forge audio files. Many criminals will forge false audio information as evidence. Therefore, audio forensics technology becomes particularly important. Audio recording device identification technology, which can verify the authenticity and uniqueness of the evidence obtained, is one of the promising branches of audio forensics technology. In this article, a novel neural-network-based framework using the device noise feature is proposed to identify the source of recording according to the device traces generated by the device during the recording. We also propose a new neural network model RARS (Recognition of Audio Recording Source based on residual neural network). The proposed framework achieves state-of-the-art performance on MOBIPHONE, the only publicly available dataset in this field. Moreover, we build a new dataset based on the latest mobile phones and tablet devices. Our method achieves good performance on both the two datasets, which proves that our model has a certain degree of reusability and robustness.

This paper deals with a score–audio music retrieval task where the aim is to find relevant audio recordings of Western classical music, given a short monophonic musical theme in symbolic notation as a query. Strategies for comparing score and audio data are often based on a common mid-level representation, such as chroma features, which capture melodic and harmonic properties. Recent studies demonstrated the effectiveness of neural networks that learn task-specific mid-level representations. Usually, such supervised learning approaches require score–audio pairs where the score's individual note events are aligned to the corresponding time positions of the audio excerpt. However, in practice, it is tedious to generate such strongly aligned training pairs. As one contribution, we show how to apply the Connectionist Temporal Classification (CTC) loss in the training procedure, which only uses weakly aligned training pairs. In such a pair, only the time positions of the beginning and end of a theme occurrence are annotated in an audio recording, rather than requiring local alignment annotations. We evaluate the resulting features in our theme retrieval scenario and show that they improve the state of the art for this task. As a main result, we demonstrate that with the CTC-based training procedure using weakly annotated data, we can achieve results almost as good as with strongly annotated data. Furthermore, we assess our chroma features in depth by inspecting their temporal smoothness or granularity as an important property and by analyzing the impact of different degrees of musical complexity on the features.

In the modern educational process music adapts to changes of everyday life and develops in the context of modern technologies. Following the needs, interests and abilities of the preschooler, preschool teachers use music through all available means of sound reproduction, combining it with their own potential and capabilities of musical expression. This paper is the result of the growing intellectual and musical potential of a preschooler who actively engages different senses in learning music: from sound to colour. The aim is to create a computer application which facilitates music learning through learning colours. The method is determined by using the application within the Max / MSP / Jitter (Max) programming language. Max is a visual programming language whose environment is based on combining objects and applications into configurations intended for music and multimedia. The aim of the paper is overcoming the basic concepts of musical literacy through using a specific colour as a carrier of organizing the music flow. The potential of this program aimed at preschool children is that learning music in the modern educational system is understood through a compromise with the development of the demands of modern technology of which music is an integral part.

This study discusses about pronunciation errors made by medical students at S&I Learning Centre. The study aims to describe the types of pronunciation errors found in students' utterances using descriptive qualitative method. The subject are medical students at S&I Learning, meanwhile the object of this exploration are students' utterances in English. The data was collected through audio recording and then those was analysed by using theory proposed by Corder by classifying errors into three types. The results show that the sorts of pronunciation errors created by medical student and 2) the cause of pronunciation errors produced by medical students. Students’ pronunciation errors are also classified into three types. Those are interference errors, intralingual errors, and developmental errors. In speaking, teacher and students had their effect to each other. The teacher has duties to assess and address the error of students by addressing the errors produced by learners permanently, particularly at the time when they drilled and practiced the correct pronunciation in speaking skill. In addition, students too should be more dynamic in surveying and correcting the pronunciation themselves. They have to create and improve their pronunciation capacity by appreciating the hypothetical about legitimate phonological. In accordance with this reason, it is necessary to apply an appropriate learning model to reduce pronunciation errors produced by students.

In this article, an artificial neural network is proposed to classify short audio sequences of squeak and rattle (S&R) noises. The aim of the classification is to see how accurately the trained classifier can recognize different types of S&R sounds. Having a high accuracy model that can recognize audible S&R noises could help to build an automatic tool able to identify unpleasant vehicle interior sounds in a matter of seconds from a short audio recording of the sounds. In this article, the training method of the classifier is proposed, and the results show that the trained model can identify various classes of S&R noises: simple (binary classification) and complex ones (multi class classification). © 2020 Institute of Noise Control Engineering.

The purpose of the article is to research and consider the general problems and perspectives of voice synthesizer development based on their cross-platform, description and analysis of the development of the author’s speech synthesis program, which can support work on the maximum number of systems capable of sound reproduction and independence of operating system.The research methods are methods of semantic analysis of the basic concepts of the considered subject area (voice synthesizers). The article considers approaches to the development of cross-platform voice synthesizers.The novelty of the research is the analysis of the mathematical apparatus that can be used for voice synthesis (vowels, different types of consonant sounds) and the description of some aspects of the development of their own author’s system that synthesizes the human voice.Conclusions. The vast majority of voice synthesis programs are designed specifically for certain types of computers or are tied to a specific operating system. The authors proposed a voice synthesizer, determined its mathematical justification and prospects for its cross-platform., Целью статьи является исследование и рассмотрение общих проблем и перспектив разработки голосовых синтезаторов на основе их кроссплатформенности, описание и анализ процессов разработки авторской программы синтеза речи, поддерживающей работу на максимальном количестве систем, способных к звуковоспроизведению и независимых от типа операционной системы.Методами исследования являются методы семантического анализа основных понятий рассматриваемой предметной области (голосовые синтезаторы). В статье исследованы подходы к разработке кроссплатформенных голосовых синтезаторов.Новизной проведенного исследования является анализ математического аппарата, применяющегося для синтеза голоса (гласных, различных типов согласных звуков), и описание некоторых аспектов разработки собственной авторской системы, синтезирующей голос человека.Выводы. Большинство программ голосового синтеза разработано специально для определенных типов компьютеров или привязано к конкретной операционной системе. В работе исследован голосовой синтезатор, определено его математическое обоснование и перспективы кроссплатформенности., Метою статті є дослідження і розгляд загальних проблем і перспектив розробки голосових синтезаторів на основі їхньої кросплатформності, опис та аналіз процесів розробки авторської програми синтезу мови, яка може підтримувати роботу на максимальній кількості систем, спроможних до звуковідтворення та незалежних від типу операційної системи.Методами дослідження є методи семантичного аналізу основних понять розглянутої предметної сфери (голосові синтезатори). У статті розглянуто підходи до розробки кросплатформних голосових синтезаторів.Новизною проведеного дослідження є аналіз математичного апарату, що може застосовуватися для синтезу голосу (голосних, різних типів приголосних звуків), та опис де­яких аспектів розробки власної авторської системи, що синтезує голос людини.Висновки. Переважна більшість програм голосового синтезу розроблена спеціально для визначених типів комп’ютерів або прив’язана до конкретної операційної системи. У роботі досліджено голосовий синтезатор, визначено його математичне обґрунтування та перспективи кросплатформності.

The aim of spatial sound or spatial audio is to reproduce the spatial information of sound, so as to recreate the desired spatial auditory events or perceptions. Recently, spatial sound becomes a hot topic in the fields of acoustics, signal processing, and communication. A series of spatial sound techniques have been developed and applied to a wide area of scientific research, engineering, and amusement. The history, principle, progress and applications of spatial sound technique are comprehensively reviewed in this article. Especially, spatial sound techniques based on different principles are united within the framework of spatial sampling and reconstruction theorem of sound field. The challenges and prospects of spatial sound are also addressed.

The subject of this research is the principle of immediacy as one of the fundamental principles of civil procedure, defining the content of proving cases heard in the order established by the Civil Procedural Code of the Russian Federation. The goal of this article consists in a thorough analysis of the application of principle of immediacy within the framework of proving, which is regulated by the norms of the civil procedural legislation. The relevance of this topic is substantiated by vast number of exceptions in application of the aforementioned procedural legal principle, including those set by the current legislation. This research employs such methods as analysis, systemic-structural, hermeneutic, and formal-legal. Based on the acquired results, the author determines a list instances in which the current legislation allows for the possibility of limiting the implementation of the principle of immediacy within the process of examination and assessment of evidence by the court. Analysis is also conducted into the limits of application of the principle of immediacy with regards to each individual case.

Click here for manuscript sample templateRecently, signal processing methods are shown to be successful while diagnosing faults in mechanical systems, using noise or vibration data. In this study, two different faulty air disc brakes; noisy and less noisy ones are investigated using Wavelet Synchrosqueezed Transform on audio recordings. The difference between two types are shown in scalogram and also verified by a quantitative measure of entropy. The audio recording has been carried out by using two identical microphones sited on the brakes via data acquisition unit at a sampling rate of 20 kHz, 16-bit resolution and these data are analyzed in MATLAB software. The average of the entropy values of faulty and non-faulty brakes were found to be 0.98 and 0.65, respectively. Therefore, it has been concluded that, the entropy could be used as a distinguishing tool to discriminate the faults.

Automatic Chord Estimation (ACE) is a fundamental task in Music Information Retrieval (MIR) and has applications in both music performance and MIR research. The task consists of segmenting a music recording or score and assigning a chord label to each segment. Although it has been a task in the annual benchmarking evaluation MIREX for over 10 years, ACE is not yet a solved problem, since performance has stagnated and modern systems have started to tune themselves to subjective training data. We propose DECIBEL, a new ACE system that exploits heterogeneous musical representations, specifically MIDI and tab files, to improve audio-based ACE methods. From an audio file and a set of MIDI and tab files corresponding to the same popular music song, DECIBEL first estimates chord sequences. For audio, state-of-the-art audio ACE methods are used. MIDI files are aligned to the audio, followed by a MIDI chord estimation step. Tab files are transformed into untimed chord sequences and then aligned to the audio. Next, DECIBEL uses data fusion to integrate all estimated chord sequences into one final output sequence. DECIBEL improves all tested state-of-the-art ACE methods by 0.5 to 13.6 percentage points. This result shows that the integration of crowd-sourced annotations from heterogeneous symbolic music representations using data fusion is a suitable strategy for addressing challenging MIR tasks such as ACE.

Disease detection is quite a difficult task and it requires medical training and experience. When it comes to heart disease diagnostics, it gets even more complex due to the additional specialization and expertise required. One of the common ways for detecting heart diseases used by heart specialists is by listening to the patient's heartbeat via a stethoscope and interpreting the sound. In this paper, we describe how we can use deep learning techniques to detect heart diseases in patients by analyzing audio recording or real time streaming of heartbeats to predict whether the patient is healthy or sick. The deep neural network model developed is a hybrid model as it combines the innate characteristics of convolutional and recurrent neural network models, which helped achieve higher accuracy in the prediction results as compared to current published results. The dataset used for training the model is a combination of two separate data science competition datasets. We used data augmentation techniques on this combined dataset to increase its size and diversity. When we tested the final model, it showed a significantly better performance as compared to the state of the art.

Communication over the phone has always been one of the most prone communication channels as it just requires a phone number to get connected and audio to communicate. Other than the already "spam" categorized calls of telecommunication companies providing offers and services, there are many other kinds of suspicious and unauthorized phone calls where the caller just stays behind the screen pretending to be a legit and an authorized person while being a non-legit one with malicious intent. People often come across the real-life incidents of such fraudulent activities including fund transfer on a call through illegal affairs or using the phone to make a threat or blackmail call to somebody. The solution for those fraudulent activity is to have software that can automatically categorize the call as non-legit or legit call and also store important points of conversations in the form of text-based notes for future reference, along with the audio recordings. The solution, therefore, is an ML based mobile application that can run in the background, record the call and analyse the conversations by using the ML algorithms to build the classification model. The classification obtained would then lead to the call being categorized as legit/spam/threat/blackmail and all the conversations would also be converted and stored as notes along with the audio recording for future reference.

Acoustic scene classification (ASC) refers to the identification of the environment in which audio excerpts have been recorded. It associates a semantic label to each audio recording. This task has recently drawn a lot of attention as a result of electronics such as smartphones, autonomous robots, or security systems acquiring the ability to perceive sounds. State-of-the-art sound scene classification heavily relies on deep neural network models. However, the complexity of these models makes them more prone to overfitting. The most widely used approach to overcome this concern is data augmentation. In this paper, we design and analyze the behavior of multiple deep learning-based acoustic scene classification systems. These systems are built following two deep convolutional neural network architectures which are defined with different characteristics. Moreover, this work deeply explores the use of Mixup data augmentation method and the effects of varying its hyperparameters. The obtained results indicate that proper tuning of Mixup hyperparameter significantly improves the classification performance, while considering the network architecture being employed.

The goal of acoustic scene classification (ASC) is to automatically classify the environment based on an audio recording of the situation. It has become a significant but difficult issue in audio processing, allowing for a vast variety of successive applications such as security, social care, and context-aware facilities. Recently, Convolutional Neural Networks (CNNs) have been broadly applied to speech audio classification, yielding promising results. In this paper, we have extracted four types of acoustic features i.e., Mel-Frequency Cepstral Coefficients (MFCCs), Linear Predictive Coding (LPC), Zero-crossing-rate (ZCR) and root mean square value (RMSV) value from audio signals for automatic scene classification. We have used Litis Rouen Dataset for CNN model training and testing. Experimental results show that the proposed method has achieved a significant accuracy of 84.15% on the Litis Rouen Dataset. Moreover, we have also compared the performance of the proposed method with other ASC techniques used in the literature.

The use of audio recordings has become a taken-for-granted approach to generating transcripts of in-depth interviewing and group discussions. In this paper we begin by describing circumstances where the use of a recorder is not, or may not be, possible, before sharing our comparative analysis of audio-recorded transcriptions and interview scripts made from notes taken during the interview (by experienced, well-trained interviewers). Our comparison shows that the data quality between audio-recorded transcripts and interview scripts written directly after the interview were comparable in the detail captured. The structures of the transcript and script were usually different because in the interview scripts, topics and ideas were grouped, rather than being in the more scattered order of the conversation in the transcripts. We suggest that in some circumstances not recording is the best approach, not ‘second best’.

Discusses the information storage work done by the York University Computer Museum(YUCoM) in 2016. YUCoM received 32 cassette tapes created at a Toronto-based computer firm Micro Computer Machines (MCM) between 1974 and 1978. These tapes once contained software and data destined for the MCM/ 70 personal computer—possibly the earliest commercially manufactured PC (1974)—and its later upgrades and refinements. Despite the luck of detailed technical information about the tape storage organization adopted by MCM, YUCoM embarked on the MCM tape recovery and analysis project because the tapes might have still contained the earliest examples of personal software written for PCs, and that could shed some new light on this software genre’s development in its preindustrial stage.1 This article chronicles the MCM tape recovery and analysis project undertaken at YUCoM.

Increasing interest in unmanned aerial vehicles (UAVs), commonly referred to as drones, has occurred in recent years. Search and rescue scenarios where humans in emergency situations need to be quickly found in difficult-toaccess areas constitute an important field of application for this technology. Drones have already been used by humanitarian organizations in countries such as Haiti and the Philippines to map areas after a natural disaster using high-resolution embedded cameras, as documented in a recent United Nations report [1]. Although research efforts have focused mostly on developing video-based solutions for this task [2], UAV-embedded audio-based localization has received relatively less attention [3]?[7]. However, UAVs equipped with a microphone array could be of critical help to localize people in emergency situations, especially when video sensors are limited by a lack of visual feedback due to bad lighting conditions (such as at night or in fog) or obstacles limiting the field of view (Figure 1).

Recent advances in music technology include practical tools for sound recording and production in school music classrooms. Secondary school music production classes allow students to make meaningful connections between school music and the music in their own lives. We offer several projects for teaching music production and sound recording; provide examples of authentic, performance-based assessments; and identify opportunities for collaboration through digital means. These projects are particularly well aligned with the United States’ Core Arts Standards related to creating music and may widen the door for students who are less interested—or less able to participate—in traditional bands, orchestras, and choirs or music appreciation electives.

This IoT based support system is an attempt on extending a contribution into lowering the increasing crime rates against children and adults with awareness and technology. The most powerful tool to this race is technology and through it, we want to provide the basic protection and safety assurance to the users. The proposed device/procedures are aimed to provide solution in case of emergencies. The plan is to merge a hardware device with the smartphones that we carry with ourselves daily. The use of the applications of that phone, without physically using it is the focal point of the project. It can be used as an advanced SOS device or a tracking device and can also we developed into a fit band.

Coding is an integral part of qualitative research for many scholars that use interview or focus group data. However, current practices in coding require transcription of audio/visual data prior to coding. Transcription before the coding process is an essential process for data analysis and even with meticulous detail, the nuances of nonverbal behavior found in audio and video data can be missed. In this article, we propose an alternative to coding with transcripts using a method called live coding which allows for simultaneous manual coding while listening or watching audio or video recording. We compared the method of live coding with transcript coding of text using focus group data from a perinatal telehealth group addressing depression. Based on the themes that emerged from analyzing the process, it is likely that live coding can be beneficial in preserving the voice of the participant especially used within focus group data. Live coding allowed us to see and hear the participants, an empowering process which allowed intent, context, and meaning of the words to be present in the results. Further study of live coding should include using digital tools for the analysis of qualitative data.

We introduce a dataset for facilitating audio-visual analysis of music performances. The dataset comprises 44 simple multi-instrument classical music pieces assembled from coordinated but separately recorded performances of individual tracks. For each piece, we provide the musical score in MIDI format, the audio recordings of the individual tracks, the audio and video recording of the assembled mixture, and ground-truth annotation files including frame-level and note-level transcriptions. We describe our methodology for the creation of the dataset, particularly highlighting our approaches for addressing the challenges involved in maintaining synchronization and expressiveness. We demonstrate the high quality of synchronization achieved with our proposed approach by comparing the dataset with existing widely-used music audio datasets. We anticipate that the dataset will be useful for the development and evaluation of existing music information retrieval (MIR) tasks, as well as for novel multi-modal tasks. We benchmark two existing MIR tasks (multi-pitch analysis and score-informed source separation) on the dataset and compare with other existing music audio datasets. Additionally, we consider two novel multi-modal MIR tasks (visually informed multi-pitch analysis and polyphonic vibrato analysis) enabled by the dataset and provide evaluation measures and baseline systems for future comparisons (from our recent work). Finally, we propose several emerging research directions that the dataset enables., 14 pages, 10 figures

Recently, end-to-end (E2E) neural text-to-speech systems, such as Tacotron2, have begun to surpass the traditional multi-stage hand-engineered systems, with both simplified system building pipelines and high-quality speech. With a unique encoder-decoder neural structure, the Tacotron2 system no longer needs separately learned text analysis front-end, duration model, acoustic model, and audio synthesis module. The key of such a system lies in the attention mechanism, which learns an alignment between the encoder and the decoder, serving as an implicit duration model bridging the text sequence and the acoustic sequence. However, attention learning suffers from low training efficiency and model instability problems, which hinder the E2E approaches from wide deployment. In this paper, we address the problems and propose a novel pre-alignment guided attention learning approach. Specifically, we inject handy prior knowledge-accurate phoneme durations-in the neural network loss function to bias the attention learning to the desired direction more accurately. The explicit time alignment between an audio recording and its corresponding phoneme sequence can be achieved by forced-alignment from an automatic speech recognizer (ASR). The experiments show that the proposed pre-alignment guided (PAG) attention approach can significantly improve training efficiency and model stability. More specifically, the PAG updated version of Tacotron2 can quickly obtain the attention alignment using only 500 (text, audio) pairs, which is apparently not possible for the original Tacotron2. A series of subjective experiments also show that the PAG-Tacotron2 approach can synthesize more stable and natural speech.

Teleconferencing has become a fast-rising communication way due to the convenience of conveying visual and auditory information. Nowadays, the efficiency of video conference including real-time, low latency and high speech quality is common, but gaps still exist between virtual presence in video conference and the real one. Immersive video conference provides more vivid sound experience and improved speech intelligibility. However, the large number of loudspeakers used in immersive sound reproduction system leads much higher computational complexity in acoustical echo cancellation. In this work, an immersive video conference implementation based on Wave Field Synthesis (WFS) technique is introduced, and an efficient frequency domain partitioned adaptive multi-channel acoustical echo cancellation algorithm based on the object-audio characteristics of WFS is used to achieve acceptable latency and computation load.

Records have become an essential source for popular music history. The article discusses the technical development and global expansion of the international recording industry from the 1890s to the 2010s. Production peaked in 1929 and started to grow again after the Second World War. As the technology developed, independent recording studios were established, and the infrastructure for making records became widely available. The article also discusses the special case of record production in the Soviet Union. In the early days of radio, stations produced music themselves, but since the 1960s, radio has mostly relied on records. Many public broadcasters still have large music archives. Television has also been important for documenting and marketing music. Since 2000, the recording industry has made a successful transition to the internet. Access to social networks has given birth to a large number of microproducers, and the role of music videos has increased.

The purpose of the article is to determine the stages of development of music broadcasting and to reveal their features in the context of the evolution of audio recording means. The methodology consists of the application of general scientific and special methods, namely: information and system approaches, terminological and historical methods, as well as methods of analysis and generalization of source information, comparison and interrelation of theory and practice. Scientific novelty. The significance of musical resources has been brought to the development of radio communications and the early stages of development of musical radio communications have been clarified in the context of the evolution by means of audio recording. Conclusions. Radio has played one of the leading roles in the field of music and, conversely, without music, radio broadcasting has not received the appropriate technical development and, accordingly, would not have had such an impact on society. Clusters of music radio recordings on the rights of subsystems are connected to the metasystem of information communications. Music publishers, who were then the most influential part of the industry, allied with the musicians. At the time of birth, radio notes were the main musical product, and songwriters were real stars. However, when the whole world began to buy records instead of music, the power from publishers and singers passed to record companies and cooperated with them performers. In the early stages of its existence, musical radio broadcasting underwent an evolutionary path from the musical telegraph (1876), the first radio shows with magnetic recording (1914), the rapid development of radio engineering and recording technology (1920–1940), and 3-minute rock ‘n’ roll from artisanal records (1950s) to «pirate music», which was broadcast from ships (1960s). The study of the historical and cultural preconditions for the formation and development of musical radio broadcasting at an early stage in the context of the evolution of audio recording allowed to identify three main stages: «search» (1870–1920), 2) «competitive» (early 1920s – second half of the 1940s). .), 3) «vinyl-tape» (first half of the 1950s – 1970s). The term «music broadcasting» is proposed to mean technological means of sound transmission to an unlimited number of listeners of musical compositions and / or other musical information on the radio, as well as wired radio networks or packet-switched networks, classifying them as «active phonograms». , broadcast) and «passive phonograms» (performed a functional role and transferred to the archive). Key words: Archive, Records, Storage, Music, Digitization, Radio.

For centuries researchers have used sound to monitor and study wildlife. Traditionally, conservationists have identified species by ear; however, it is now common to deploy audio recording technology to monitor animal and ecosystem sounds. Animals use sound for communication, mating, navigation and territorial defence. Animal sounds provide valuable information and help conservationists to quantify biodiversity. Acoustic monitoring has grown in popularity due to the availability of diverse sensor types which include camera traps, portable acoustic sensors, passive acoustic sensors, and even smartphones. Passive acoustic sensors are easy to deploy and can be left running for long durations to provide insights on habitat and the sounds made by animals and illegal activity. While this technology brings enormous benefits, the amount of data that is generated makes processing a time-consuming process for conservationists. Consequently, there is interest among conservationists to automatically process acoustic data to help speed up biodiversity assessments. Processing these large data sources and extracting relevant sounds from background noise introduces significant challenges. In this paper we outline an approach for achieving this using state of the art in machine learning to automatically extract features from time-series audio signals and modelling deep learning models to classify different bird species based on the sounds they make. The acquired bird songs are processed using mel-frequency cepstrum (MFC) to extract features which are later classified using a multilayer perceptron (MLP). Our proposed method achieved promising results with 0.74 sensitivity, 0.92 specificity and an accuracy of 0.74.

Speaker counting is the task of estimating the number of people that are simultaneously speaking in an audio recording. For several audio processing tasks such as speaker diarization, separation, localization and tracking, knowing the number of speakers at each timestep is a prerequisite, or at least it can be a strong advantage, in addition to enabling a low latency processing. For that purpose, we address the speaker counting problem with a multichannel convolutional recurrent neural network which produces an estimation at a short-term frame resolution. We trained the network to predict up to 5 concurrent speakers in a multichannel mixture, with simulated data including many different conditions in terms of source and microphone positions, reverberation, and noise. The network can predict the number of speakers with good accuracy at frame resolution., 5 pages, 1 figure

Spherical harmonic analysis has been a widely used approach for spatial audio processing in recent years. Among all applications that benefit from spatial processing, spatial Active Noise Control (ANC) remains unique with its requirement for open spherical microphone arrays to record the residual sound field throughout the continuous region. Ideally, a low delay spherical harmonic recording algorithm for open spherical microphone arrays is desired for real-time spatial ANC systems. Currently, frequency domain algorithms for spherical harmonic decomposition of microphone array recordings are applied in a spatial ANC system. However, a Short Time Fourier Transform is required, which introduces undesirable system delay for ANC systems. In this paper, we develop a time domain spherical harmonic decomposition algorithm for the application of spatial audio recording mainly with benefit to ANC with an open spherical microphone array. Microphone signals are processed by a series of pre-designed finite impulse response (FIR) filters to obtain a set of time domain spherical harmonic coefficients. The time domain coefficients contain the continuous spatial information of the residual sound field. We corroborate the time domain algorithm with a numerical simulation of a fourth order system, and show the proposed method to have lower delay than existing approaches.

The continuing effect of COVID-19 pulmonary infection has highlighted the importance of machine-aided diagnosis for its initial symptoms such as fever, dry cough, fatigue, and dyspnea. This paper attempts to address the respiratory-related symptoms, using a low power scalable software and hardware framework. We propose CoughNet, a flexible low power CNN-LSTM processor that can take audio recordings as input to detect cough sounds in audio recordings. We analyze the three different publicly available datasets and use those as part of our evaluation to detect cough sound in audio recordings. We perform windowing and hyperparameter optimization on the software side with regard to fitting the network architecture to the hardware system. A scalable hardware prototype is designed to handle different numbers of processing engines and flexible bitwidth using Verilog HDL on Xilinx Kintex-7 160t FPGA. The proposed implementation of hardware has a low power consumption of o 290 mW and energy consumption of 2 mJ which is about $99 \times $ less compared to the state-of-the-art implementation.

International audience; The slow but steady shift away from printed text into digital media has not yet modified the working habits of chamber music practitioners. If most instrumentalists still heavily rely on printed scores, audiences increasingly access notated music online, with printed scores synced to an audio recording on youtube for instance. This paper proposes to guide the listener and/or the performer with a cursor scrolling on the page with INScore, in order to examine the consequences of representing time in this way as opposed to traditional bars and beats notation. In addition to its score following interest for pedagogy and analysis, the networking possibilities of today's ubiquitous technologies reveal interesting potentials for works in which the presence of a conductor is required for synchronization between performers and/or with fixed media (film or tape). A Raspberry Pi-embedded prototype for animated/distributed notation is presented here as a score player (such as the Decibel ScorePlayer, or SmartVox), in order to send and synchronize mp4 scores to any browser capable device connected to the same WIFI network. The corpus will concern pieces edited at BabelScores, an online library for contemporary classical music. The BabelScores pdf works, composed in standard engraving softwares, will be animated using INScore and video editors, in order to find strategies for animation or dynamic display of the unfolding of time, originally represented statically on the page.; L'abandon lent mais certain du texte imprimé au profit des médias numériques n'a pas encore modifié les habitudes de travail des instrumentistes en musique de chambre. Si la plupart s'appuie encore fortement sur les partitions imprimées, le public a de plus en plus accès aux partitions en ligne, synchronisées sur un enregistrement audio sur youtube par exemple. Cet article propose de guider l'auditeur et/ou l'interprète à l'aide d'un curseur défilant sur la page (avec INScore), afin d'examiner ce qu'implique de représenter le temps de cette manière, par opposition à la notation traditionnelle des mesures rythmiques du solfège. Outre son intérêt pour la pédagogie et l'analyse, les possibilités de mise en réseau des technologies omniprésentes d'aujourd'hui révèlent des potentiels intéressants pour des œuvres dans lesquelles la présence d'un chef d'orchestre est requise pour la synchronisation entre interprètes et/ou avec des supports fixes (film ou bande). Un prototype pour la notation animée/distribuée intégré sur Raspberry Pi est présenté ici comme lecteur de partition (tel que le Decibel ScorePlayer ou SmartVox), afin d’envoyer et de synchroniser les partitions mp4 à n’importe quel appareil compatible avec un navigateur connecté au même réseau WIFI. Le corpus concernera des pièces éditées sur BabelScores, une bibliothèque en ligne de musique classique contemporaine. Les œuvres pdf de BabelScores, composées de logiciels de gravure standard, seront animées à l'aide d'éditeurs INScore et vidéo, afin de trouver des stratégies d'animation ou d'affichage dynamique du déroulement du temps, initialement représentées de manière statique sur la page.

Speech Activity Detection (SAD), locating speech segments within an audio recording, is a main part of most speech technology applications. Robust SAD is usually more difficult in noisy conditions with varying signal-to-noise ratios (SNR). The Fearless Steps challenge has recently provided such data from the NASA Apollo-11 mission for different speech processing tasks including SAD. Most audio recordings are degraded by different kinds and levels of noise varying within and between channels. This paper describes the EML online algorithm for the most recent phase of this challenge. The proposed algorithm can be trained both in a supervised and unsupervised manner and assigns speech and non-speech labels at runtime approximately every 0.1 sec. The experimental results show a competitive accuracy on both development and evaluation datasets with a real-time factor of about 0.002 using a single CPU machine.

A common problem in audio forensics is the difficulty to authenticate an audio recording. You can have an argument against any audio probe. In this paper we propose a system that can take audio probes which can be used with success in a court. The recording will be bulletproof due to the use of a reference protocol based on an external entity, the company that provides the recording system. Using an external cloud-based application, for the recorded person and for the recording institution, the resulted audio-recording will be 100% authenticated in terms of integrity check. All editing by means of deleting/inserting audio fragment longer than 5 ms will be detected.

RESUMO: Um grande número de estudos sobre processos de escritura em tempo real usa métodos e técnicas para coleta de dados, sob condições metodológicas experimentais. Este artigo apresenta um sistema de captura multimodal (visual, sonora e escrita), capaz de oferecer ao pesquisador informações simultâneas do processo de escritura no tempo e espaço da sala de aula (contexto ecológico e didático). O registro visual é feito através de câmeras de vídeo, capturando o contexto da sala de aula e a interação entre alunos e professor; o registro sonoro é feito através de gravadores digitais e microfones, capturando a fala espontânea dos alunos escreventes e o diálogo entre os participantes; o registro escrito é feito através de uma caneta inteligente e do programa HandSpy, capturando o traço da tinta na folha de papel. Descrevemos cada um desses instrumentos, o modo como eles são utilizados e a técnica de sincronização, gerando uma única mídia (filme-sincronizado). Para ilustrar como funciona o Sistema Ramos, sua importância e as vantagens que oferece para entendermos melhor o processo de produção textual (gênese textual) em contexto didático e ecológico, apresentamos uma análise qualitativa e microgenética de pausas espontâneas feitas por uma dupla de alunos (9 anos) enquanto escrevia uma história inventada.

Audio has been increasingly adopted as a sensing modality in a variety of human-centered mobile applications and in smart assistants in the home. Although acoustic features can capture complex semantic information about human activities and context, continuous audio recording often poses significant privacy concerns. An intuitive way to reduce privacy concerns is to degrade audio quality such that speech and other relevant acoustic markers become unintelligible, but this often comes at the cost of activity recognition performance. In this paper, we employ a mixed-methods approach to characterize this balance. We first conduct an online survey with 266 participants to capture their perception of privacy qualitatively and quantitatively with degraded audio. Given our findings that privacy concerns can be significantly reduced at high levels of audio degradation, we then investigate how intentional degradation of audio frames can affect the recognition results of the target classes while maintaining effective privacy mitigation. Our results indicate that degradation of audio frames can leave minimal effects for audio recognition using frame-level features. Furthermore, degradation of audio frames can hurt the performance to some extend for audio recognition using segment-level features, though the usage of such features may still yield superior recognition performance. Given the different requirements on privacy mitigation and recognition performance for different sensing purposes, such trade-offs need to be balanced in actual implementations.

To enrich the form of film and television production, improve the level of film and television production, and satisfy the film-watching experiences of audiences, based on Virtual Reality (VR) and the Internet of Things (IoT) technology, with the help of S3 Studio Max and Photoshop software, a VR film-watching system is built, which realizes the interaction with users on different devices through somatosensory interaction sensors. In addition, by utilizing Twirling720, the panoramic sound recording is achieved. Through this system, a smart IoT platform between users, films, and devices is built. Finally, this platform is utilized to produce the film and television work Van Gogh in Dream, which is evaluated and analyzed through questionnaires. The results show that the technology system of this set of film and television production is complete, and the production level of film and television works have been significantly improved. The audience recognition of film and television production based on this technology is 55%, and the impression evaluation is over 56%. However, knowledge acquisition is only 20%, and historical understanding is above 50%. These dimensions show that compared with traditional film production, artificial intelligence films can bring a better experience to audiences, but knowledge acquisition is less. Therefore, professional knowledge will be improved at the later stage. The above results provide a theoretical basis for the application of artificial intelligence technology in film production and production mode.

Various open-source toolkits exist for speech recognition and speech processing. These toolkits have brought a great benefit to the research community, i.e. speeding up research. Yet, no such freely available toolkit exists for automatic affect recognition from speech. We herein introduce a novel open-source affect and emotion recognition engine, which integrates all necessary components in one highly efficient software package. The components include audio recording and audio file reading, state-of-the-art paralinguistic feature extraction and plugable classification modules. In this paper we introduce the engine and extensive baseline results. Pre-trained models for four affect recognition tasks are included in the openEAR distribution. The engine is tailored for multi-threaded, incremental on-line processing of live input in real-time, however it can also be used for batch processing of databases.

The increase in low cost digital audio recorders, mobile applications and accelerating growth of the audio-based Internet of Things (IoT) has initiated ease in obtaining speech and audio data which will be later used for the purpose of identifying human traits, implementing voice authentication system and development of voice based embedded systems. On the other hand, the availability of free advanced mobile applications and audio editing software like Adobe, Audition CC, etc. enabling people to edit easily the meaningful content of audio recordings data for getting benefit from e-services or producing it in a courtroom for the purpose digital proof. Perhaps, most of the people do it for fun as well as the strong intention of hiding reality present. Moreover, the audio recording captured in a real-life today does not contain digital watermarking and signature content for authentication because of expensive procedure. Therefore, in recent years the researches focused more on developing active audio forgery detection techniques for copy-move and audio splicing forgeries to authenticate and verify for its genuineness. In this paper, put an effort to describe past and present developments in audio copy-move and audio splicing forgery techniques. The paper also presents an overview of audio forgeries, its classification, various post-processing operations used and available audio dataset for forgery sample preparation for benchmark testing.

Current automatic cough counting systems can determine how many coughs are present in an audio recording. However, they cannot determine who produced the cough. This limits their usefulness as most systems are deployed in locations with multiple people (i.e., a smart home device in a four-person home). Previous models trained solely on speech performed reasonably well on forced coughs [1]. By incorporating coughs into the training data, the model performance should improve. However, since limited natural cough data exists, training on coughs can lead to model overfitting. In this work, we overcome this problem by using multitask learning, where the second task is speaker verification. Our model achieves 82.15% classification accuracy amongst four users on a natural, in-the-wild cough dataset, outperforming human evaluators on average by 9.82%.

We can naturally observe in group dynamics when some people speak too much, others too little, or when one person consistently speaks over another, however, we often have little objective data to support such assertions. Even when such data can be collected, it is not often processed quickly enough to provide quality feedback to affect behavior. We developed the conversation moderator to provide feedback on the engagement of people speaking in the meeting through a friendly and intuitive user interface. The application utilizes machine learning-based speaker diarization to parse the audio recording into separate individuals, and the results are presented in a variety of visual representations. This includes total speaking time for each person, a scrollable piano chart indicating the time course of each speaker throughout the conversation, and additional statistics on speaking intervals. Diarization was performed with CMU SPHINX for MFCC conversion to extract audio features relevant for speech recognition and the LIUM toolkit to perform the necessary clustering and sequence estimation to distinguish each speaker. This application provides unbiased information on speakers with the goal of facilitating group discussions and aiding in the study of human social interaction.

Manipulating speech audio recordings through splicing is a task within everyone’s reach. Indeed, it is very easy to collect through social media multiple audio recordings from well-known public figures (e.g., actors, politicians, etc.). These can be cut into smaller excerpts that can be concatenated in order to generate new audio content. As a fake speech from a famous person can be used for fake news spreading and negatively impact on the society, the ability of detecting whether a speech recording has been manipulated is a task of great interest in the forensics community. In this work, we focus on speech audio splicing detection and localization. We leverage the idea that distinct recordings may be acquired in different environments, which are typically characterized by distinctive reverberation cues. Exploiting this property, our method estimates inconsistencies in the reverberation time throughout a speech recording. If reverberation inconsistencies are detected, the audio track is tagged as manipulated and the splicing point time instant is estimated.

Последние нововведения в уголовном процессе затронули вопросы процессуальной фиксации введено аудиопротоколирование судебного заседания. В статье исследуются проблемы введения аудиозаписи в уголовном процессе, ее преимущества и недостатки. Авторами сделаны выводы о необходимости сохранения письменной формы протокола судебного заседания и внесении некоторых изменений в Уголовно-процессуальный кодекс РФ., The latest innovations in criminal proceedings touched upon the issues of procedural fi xation audio recording of court sessions was introduced. The article investigates the problems of introduction of audio recording in criminal proceedings, its advantages and disadvantages. The authors draw conclusions about the need to preserve the written form of the minutes of the court session and to make some changes to the Code of Criminal Procedure of the Russian Federation., Научное мнение. Экономические, юридические и социологические науки, Выпуск 2 2020, Pages 45-49

Developments in artificial intelligence can be leveraged to support the diagnosis of degenerative disorders, such as epilepsy and Parkinson’s disease. This study aims to provide a software solution, focused initially towards Parkinson’s disease, which can positively impact medical practice surrounding degenerative diagnoses. Through the use of a dataset containing numerical data representing acoustic features extracted from an audio recording of an individual, it is determined if a neural approach can provide an improvement over previous results in the area. This is achieved through the implementation of a feedforward neural network and a layer recurrent neural network. By comparison with the state-of-the-art, a Bayesian approach providing a classification accuracy benchmark of 87.1%, it is found that the implemented neural networks are capable of average accuracy of 96%, highlighting improved accuracy for the classification process. The solution is capable of supporting the diagnosis of Parkinson’s disease in an advisory capacity and is envisioned to inform the process of referral through general practice.