Search
Search results for: SPEECH SIGNAL
-
Weakly-Supervised Word-Level Pronunciation Error Detection in Non-Native English Speech
PublicationWe propose a weakly-supervised model for word-level mispronunciation detection in non-native (L2) English speech. To train this model, phonetically transcribed L2 speech is not required and we only need to mark mispronounced words. The lack of phonetic transcriptions for L2 speech means that the model has to learn only from a weak signal of word-level mispronunciations. Because of that and due to the limited amount of mispronounced...
-
Metoda i algorytmy modyfikacji sygnału do celu wspomagania rozumienia mowy przez osoby z pogorszoną rozdzielczością czasową słuchu
PublicationPrzedmiotem badań przeprowadzonych w ramach rozprawy są metody modyfikacji czasu trwania sygnału (ang. Time Scale Modification –TSM) mowy operujące w czasie rzeczywistym oraz ocena ich wpływu na rozumienie wypowiedzi przez osoby z pogorszoną rozdzielczością czasową słuchu. Pogorszona rozdzielczość słuchu jest jednym z symptomów związanych z ośrodkowymi zaburzeniami słuchu (ang. Cetnral Auditory Processing Disorder – CAPD). W odróżnieniu...
-
Computer-assisted pronunciation training—Speech synthesis is almost all you need
PublicationThe research community has long studied computer-assisted pronunciation training (CAPT) methods in non-native speech. Researchers focused on studying various model architectures, such as Bayesian networks and deep learning methods, as well as on the analysis of different representations of the speech signal. Despite significant progress in recent years, existing CAPT methods are not able to detect pronunciation errors with high...
-
Human voice modification using instantaneous complex frequency
PublicationThe paper presents the possibilities of changing human voice by modifying instantaneous complex frequency (ICF) of the speech signal. The proposed method provides a flexible way of altering voice without the necessity of finding fundamental frequency and formants' positions or detecting voiced and unvoiced fragments of speech. The algorithm is simple and fast. Apart from ICF it uses signal factorization into two factors: one fully...
-
AUTOMATYCZNA KLASYFIKACJA MOWY PATOLOGICZNEJ
PublicationAplikacja przedstawiona w niniejszym rozdziale służy do automatycznego wykrywania mowy patologicznej na podstawie bazy nagrań. W pierwszej kolejności przedstawiono założenia leżące u podstaw przeprowadzonych badan wraz z wyborem bazy mowy patologicznej. Zaprezentowano również zastosowane algorytmy oraz cechy sygnału mowy, które pozwalają odróżnić mowę niezaburzoną od mowy patologicznej. Wytrenowane sieci neuronowe zostały następnie...
-
Audio-visual aspect of the Lombard effect and comparison with recordings depicting emotional states.
PublicationIn this paper an analysis of audio-visual recordings of the Lombard effect is shown. First, audio signal is analyzed indicating the presence of this phenomenon in the recorded sessions. The principal aim, however, was to discuss problems related to extracting differences caused by the Lombard effect, present in the video , i.e. visible as tension and work of facial muscles aligned to an increase in the intensity of the articulated...
-
A low complexity double-talk detector based on the signal envelope
PublicationA new algorithm for double-talk detection, intended for use in the acoustic echo canceller for voice communication applications, is proposed. The communication system developed by the authors required the use of a double-talk detection algorithm with low complexity and good accuracy. The authors propose an approach to doubletalk detection based on the signal envelopes. For each of three signals: the far-end speech, the microphone...
-
Marking the Allophones Boundaries Based on the DTW Algorithm
PublicationThe paper presents an approach to marking the boundaries of allophones in the speech signal based on the Dynamic Time Warping (DTW) algorithm. Setting and marking of allophones boundaries in continuous speech is a difficult issue due to the mutual influence of adjacent phonemes on each other. It is this neighborhood on the one hand that creates variants of phonemes that is allophones, and on the other hand it affects that the border...
-
Enhanced voice user interface employing spatial filtration of signals from acoustic vector sensor
PublicationSpatial filtration of sound is introduced to enhance speech recognition accuracy in noisy conditions. An acoustic vector sensor (AVS) is employed. The signals from the AVS probe are processed in order to attenuate the surrounding noise. As a result the signal to noise ratio is increased. An experiment is featured in which speech signals are disturbed by babble noise. The signals before and after spatial filtration are processed...
-
Quality Evaluation of Speech Transmission via Two-way BPL-PLC Voice Communication System in an Underground Mine
PublicationIn order to design a stable and reliable voice communication system, it is essential to know how many resources are necessary for conveying quality content. These parameters may include objective quality of service (QoS) metrics, such as: available bandwidth, bit error rate (BER), delay, latency as well as subjective quality of experience (QoE) related to user expectations. QoE is expressed as clarity of speech and the ability...
-
Analysis of allophones based on audio signal recordings and parameterization
PublicationThe aim of this study is to develop an allophonic description of English plosive consonants based on recordings of 600 specially selected words. Allophonic variations addressed in the study may have two sources: positional and contextual. The former one depends on the syllabic or prosodic position in which a particular phoneme occurs. Contextual allophony is conditioned by the local phonetic environment. Co-articulation overlapping...
-
Estimation of time-frequency complex phase-based speech attributes using narrow band filter banks
PublicationIn this paper, we present nonlinear estimators of nonstationary and multicomponent signal attributes (parameters, properties) which are instantaneous frequency, spectral (or group) delay, and chirp-rate (also known as instantaneous frequency slope). We estimate all of these distributions in the time-frequency domain using both finite and infinite impulse response (FIR and IIR) narrow band filers for speech analysis. Then, we present...
-
Strategie treningu neuronowego estymatora częstotliwości tonu krtaniowego z użyciem generatora syntetycznych samogłosek
PublicationW wielu zastosowaniach telekomunikacyjnych pojawia się problem przetwarzania lub analizy sygnału mowy, w ramach którego, często w obszarze podstawowych algorytmów, stosuje się estymator częstotliwości tonu krtaniowego. Estymator rozpatrywany w tej pracy bazuje na neuronowym klasyfikatorze podejmującym decyzje na podstawie częstotliwości oraz mocy chwilowej wyznaczanych w podpasmach analizowanego sygnału mowy. W pracy rozważamy...
-
Noise profiling for speech enhancement employing machine learning models
PublicationThis paper aims to propose a noise profiling method that can be performed in near real-time based on machine learning (ML). To address challenges related to noise profiling effectively, we start with a critical review of the literature background. Then, we outline the experiment performed consisting of two parts. The first part concerns the noise recognition model built upon several baseline classifiers and noise signal features...
-
WYKORZYSTANIE SIECI NEURONOWYCH DO SYNTEZY MOWY WYRAŻAJĄCEJ EMOCJE
PublicationW niniejszym artykule przedstawiono analizę rozwiązań do rozpoznawania emocji opartych na mowie i możliwości ich wykorzystania w syntezie mowy z emocjami, wykorzystując do tego celu sieci neuronowe. Przedstawiono aktualne rozwiązania dotyczące rozpoznawania emocji w mowie i metod syntezy mowy za pomocą sieci neuronowych. Obecnie obserwuje się znaczny wzrost zainteresowania i wykorzystania uczenia głębokiego w aplikacjach związanych...
-
Rediscovering Automatic Detection of Stuttering and Its Subclasses through Machine Learning—The Impact of Changing Deep Model Architecture and Amount of Data in the Training Set
PublicationThis work deals with automatically detecting stuttering and its subclasses. An effective classification of stuttering along with its subclasses could find wide application in determining the severity of stuttering by speech therapists, preliminary patient diagnosis, and enabling communication with the previously mentioned voice assistants. The first part of this work provides an overview of examples of classical and deep learning...
-
Quality Evaluation of Novel DTD Algorithm Based on Audio Watermarking
PublicationEcho cancellers typically employ a doubletalk detection (DTD) algorithm in order to keep the adaptive filter from diverging in the presence of near-end speech signal or other disruptive sounds in the microphone signal. A novel doubletalk detection algorithm based on techniques similar to those used for audio signal watermarking was introduced by the authors. The application of the described DTD algorithm within acoustic echo cancellation...
-
Variable Ratio Sample Rate Conversion Based on Fractional Delay Filter
PublicationIn this paper a sample rate conversion algorithm which allows for continuously changing resampling ratio has been presented. The proposed implementation is based on a variable fractional delay filter which is implemented by means of a Farrow structure. Coefficients of this structure are computed on the basis of fractional delay filters which are designed using the offset window method. The proposed approach allows us to freely...
-
A Comparison of STI Measured by Direct and Indirect Methods for Interiors Coupled with Sound Reinforcement Systems
PublicationThis paper presents a comparison of STI (Speech Transmission Index) coefficient measurement results carried out by direct and indirect methods. First, acoustic parameters important in the context of public address and sound reinforcement systems are recalled. A measurement methodology is presented that employs various test signals to determine impulse responses. The process of evaluating sound system performance, signals enabling...
-
Analysis-by-synthesis paradigm evolved into a new concept
PublicationThis work aims at showing how the well-known analysis-by-synthesis paradigm has recently been evolved into a new concept. However, in contrast to the original idea stating that the created sound should not fail to pass the foolproof synthesis test, the recent development is a consequence of the need to create new data. Deep learning models are greedy algorithms requiring a vast amount of data that, in addition, should be correctly...
-
Intra-subject class-incremental deep learning approach for EEG-based imagined speech recognition
PublicationBrain–computer interfaces (BCIs) aim to decode brain signals and transform them into commands for device operation. The present study aimed to decode the brain activity during imagined speech. The BCI must identify imagined words within a given vocabulary and thus perform the requested action. A possible scenario when using this approach is the gradual addition of new words to the vocabulary using incremental learning methods....
-
Selection of Features for Multimodal Vocalic Segments Classification
PublicationEnglish speech recognition experiments are presented employing both: audio signal and Facial Motion Capture (FMC) recordings. The principal aim of the study was to evaluate the influence of feature vector dimension reduction for the accuracy of vocalic segments classification employing neural networks. Several parameter reduction strategies were adopted, namely: Extremely Randomized Trees, Principal Component Analysis and Recursive...
-
Comparative analysis of various transformation techniques for voiceless consonants modeling
PublicationIn this paper, a comparison of various transformation techniques, namely Discrete Fourier Transform (DFT), Discrete Cosine Transform (DCT) and Discrete Walsh Hadamard Transform (DWHT) are performed in the context of their application to voiceless consonant modeling. Speech features based on these transformation techniques are extracted. These features are mean and derivative values of cepstrum coefficients, derived from each transformation....
-
A Novel Approach to the Assessment of Cough Incidence
PublicationIn this paper we consider the problem of identication of cough events in patients suffering from chronic respiratory diseases. The information about frequency of cough events is necessary to medical treatment. The proposed approach is based on bidirectional processing of a measured vibration signal - cough events are localized by combining the results of forward-time and backward-time analysis. The signal is at rst transformed...
-
Determining Pronunciation Differences in English Allophones Utilizing Audio Signal Parameterization
PublicationAn allophonic description of English plosive consonants, based on audio-visual recordings of 600 specially selected words, was developed. First, several speakers were recorded while reading words from a teleprompter. Then, every word was played back from the previously recorded sample read by a phonology expert and each examined speaker repeated a particular word trying to imitate correct pronunciation. The next step consisted...
-
Transfer learning in imagined speech EEG-based BCIs
PublicationThe Brain–Computer Interfaces (BCI) based on electroencephalograms (EEG) are systems which aim is to provide a communication channel to any person with a computer, initially it was proposed to aid people with disabilities, but actually wider applications have been proposed. These devices allow to send messages or to control devices using the brain signals. There are different neuro-paradigms which evoke brain signals of interest...
-
Playback detection using machine learning with spectrogram features approach
PublicationThis paper presents 2D image processing approach to playback detection in automatic speaker verification (ASV) systems using spectrograms as speech signal representation. Three feature extraction and classification methods: histograms of oriented gradients (HOG) with support vector machines (SVM), HAAR wavelets with AdaBoost classifier and deep convolutional neural networks (CNN) were compared on different data partitions in respect...
-
Chirp Rate and Instantaneous Frequency Estimation: Application to Recursive Vertical Synchrosqueezing
PublicationThis letter introduces new chirp rate and instantaneous frequency estimators designed for frequency-modulated signals. These estimators are first investigated from a deterministic point of view, then compared together in terms of statistical efficiency. They are also used to design new recursive versions of the vertically synchrosqueezed short-time Fourier transform, using a previously published method (D. Fourer, F. Auger, and...
-
Ultrawideband transmission in physical channels: a broadband interference view
PublicationThe superposition of multipath components (MPC) of an emitted wave, formed by reflections from limiting surfaces and obstacles in the propagation area, strongly affects communication signals. In the case of modern wideband systems, the effect should be seen as a broadband counterpart of classical interference which is the cause of fading in narrowband systems. This paper shows that in wideband communications, the time- and frequency-domain...
-
Contactless hearing aid designed for infants
PublicationIt is a well known fact that language development through home intervention for a hearing-impaired infant should start in the early months of a newborn baby's life. The aim of this paper is to present a concept of a contactless digital hearing aid designed especially for infants. In contrast to all typical wearable hearing aid solutions (ITC, ITE, BTE), the proposed device is mounted in the infant's bed with any parts of its set-up...
-
Detection of dialogue in movie soundtrack for speech intelligibility enhancement
PublicationA method for detecting dialogue in 5.1 movie soundtrack based on interchannel spectral disparity is presented. The front channel signals (left, right, center) are analyzed in the frequency domain. The selected partials in the center channel signal, which yield high disparity with left and right channels, are detected as dialogue. Subsequently, the dialogue frequency components are boosted to achieve increased dialogue intelligibility....
-
Study Analysis of Transmission Efficiency in DAB+ Broadcasting System
PublicationDAB+ is a very innovative and universal multimedia broadcasting system. Thanks to its updated multimedia technologies and metadata options, digital radio keeps pace with changing consumer expectations and the impact of media convergence. Broadcasting analog and digital radio services does vary, concerning devices on both transmitting and receiving side, as well as content processing mechanisms. However, the biggest difference is...
-
Voiceless Stop Consonant Modelling and Synthesis Framework Based on MISO Dynamic System
PublicationA voiceless stop consonant phoneme modelling and synthesis framework based on a phoneme modelling in low-frequency range and high-frequency range separately is proposed. The phoneme signal is decomposed into the sums of simpler basic components and described as the output of a linear multiple-input and single-output (MISO) system. The impulse response of each channel is a third order quasi-polynomial. Using this framework, the...
-
Analysis of 2D Feature Spaces for Deep Learning-based Speech Recognition
Publicationconvolutional neural network (CNN) which is a class of deep, feed-forward artificial neural network. We decided to analyze audio signal feature maps, namely spectrograms, linear and Mel-scale cepstrograms, and chromagrams. The choice was made upon the fact that CNN performs well in 2D data-oriented processing contexts. Feature maps were employed in the Lithuanian word recognition task. The spectral analysis led to the highest word...
-
Audio Content and Crowdsourcing: A Subjective Quality Evaluation of Radio Programs Streamed Online
PublicationRadio broadcasting has been present in our lives for over 100 years. The transmission of speech and music signals accompanies us from an early age. Broadcasts provide the latest information from home and abroad. They also shape musical tastes and allow many artists to share their creativity. Modern distribution involves transmission over a number of terrestrial systems. The most popular are analog FM (Frequency Modulation) and...
-
Speech Analytics Based on Machine Learning
PublicationIn this chapter, the process of speech data preparation for machine learning is discussed in detail. Examples of speech analytics methods applied to phonemes and allophones are shown. Further, an approach to automatic phoneme recognition involving optimized parametrization and a classifier belonging to machine learning algorithms is discussed. Feature vectors are built on the basis of descriptors coming from the music information...
-
Interpretable Deep Learning Model for the Detection and Reconstruction of Dysarthric Speech
PublicationWe present a novel deep learning model for the detection and reconstruction of dysarthric speech. We train the model with a multi-task learning technique to jointly solve dysarthria detection and speech reconstruction tasks. The model key feature is a low-dimensional latent space that is meant to encode the properties of dysarthric speech. It is commonly believed that neural networks are black boxes that solve problems but do not...
-
Introduction to the special issue on machine learning in acoustics
PublicationWhen we started our Call for Papers for a Special Issue on “Machine Learning in Acoustics” in the Journal of the Acoustical Society of America, our ambition was to invite papers in which machine learning was applied to all acoustics areas. They were listed, but not limited to, as follows: • Music and synthesis analysis • Music sentiment analysis • Music perception • Intelligent music recognition • Musical source separation • Singing...
-
Applying the Lombard Effect to Speech-in-Noise Communication
PublicationThis study explored how the Lombard effect, a natural or artificial increase in speech loudness in noisy environments, can improve speech-in-noise communication. This study consisted of several experiments that measured the impact of different types of noise on synthesizing the Lombard effect. The main steps were as follows: first, a dataset of speech samples with and without the Lombard effect was collected in a controlled setting;...
-
Constructing a Dataset of Speech Recordingswith Lombard Effect
PublicationThepurpose of therecordings was to create a speech corpus based on the ISLEdataset, extended with video and Lombard speech. Selected from a set of 165sentences, 10, evaluatedas having thehighest possibility to occur in the context ofthe Lombard effect,were repeated in the presence of the so-called babble speech to obtain Lombard speech features. Altogether,15speakers were recorded, and speech parameterswere...
-
Orken Mamyrbayev Professor
People1. Education: Higher. In 2001, graduated from the Abay Almaty State University (now Abay Kazakh National Pedagogical University), in the specialty: Computer science and computerization manager. 2. Academic degree: Ph.D. in the specialty "6D070300-Information systems". The dissertation was defended in 2014 on the topic: "Kazakh soileulerin tanudyn kupmodaldy zhuyesin kuru". Under my supervision, 16 masters, 1 dissertation...
-
Acoustic Sensing Analytics Applied to Speech in Reverberation Conditions
PublicationThe paper aims to discuss a case study of sensing analytics and technology in acoustics when applied to reverberation conditions. Reverberation is one of the issues that makes speech in indoor spaces challenging to understand. This problem is particularly critical in large spaces with few absorbing or diffusing surfaces. One of the natural remedies to improve speech intelligibility in such conditions may be achieved through speaking...
-
Optimizing Medical Personnel Speech Recognition Models Using Speech Synthesis and Reinforcement Learning
PublicationText-to-Speech synthesis (TTS) can be used to generate training data for building Automatic Speech Recognition models (ASR). Access to medical speech data is because it is sensitive data that is difficult to obtain for privacy reasons; TTS can help expand the data set. Speech can be synthesized by mimicking different accents, dialects, and speaking styles that may occur in a medical language. Reinforcement Learning (RL), in the...
-
A survey of automatic speech recognition deep models performance for Polish medical terms
PublicationAmong the numerous applications of speech-to-text technology is the support of documentation created by medical personnel. There are many available speech recognition systems for doctors. Their effectiveness in languages such as Polish should be verified. In connection with our project in this field, we decided to check how well the popular speech recognition systems work, employing models trained for the general Polish language....
-
Material for Automatic Phonetic Transcription of Speech Recorded in Various Conditions
PublicationAutomatic speech recognition (ASR) is under constant development, especially in cases when speech is casually produced or it is acquired in various environment conditions, or in the presence of background noise. Phonetic transcription is an important step in the process of full speech recognition and is discussed in the presented work as the main focus in this process. ASR is widely implemented in mobile devices technology, but...
-
Automatic music signal mixing system based on one-dimensional Wave-U-Net autoencoders
PublicationThe purpose of this paper is to show a music mixing system that is capable of automatically mixing separate raw recordings with good quality regardless of the music genre. This work recalls selected methods for automatic audio mixing first. Then, a novel deep model based on one-dimensional Wave-U-Net autoencoders is proposed for automatic music mixing. The model is trained on a custom-prepared database. Mixes created using the...
-
Mowa nienawiści (hate speech) a odpowiedzialność dostawców usług internetowych w orzecznictwie sądów europejskich
PublicationThe article analyses the phenomenon of hate speech in the Internet contrasted with the problem of responsability of Internet Service Providers for cases of such abuses of freedom of expression. The text provides an analysis of jurisprudence of two European Courts. On the one hand it presents the position of the European Court of Human Rights on the problem of hate speech: its definition and the liability for it as an exception...
-
Automated detection of pronunciation errors in non-native English speech employing deep learning
PublicationDespite significant advances in recent years, the existing Computer-Assisted Pronunciation Training (CAPT) methods detect pronunciation errors with a relatively low accuracy (precision of 60% at 40%-80% recall). This Ph.D. work proposes novel deep learning methods for detecting pronunciation errors in non-native (L2) English speech, outperforming the state-of-the-art method in AUC metric (Area under the Curve) by 41%, i.e., from...
-
Corrupted speech intelligibility improvement using adaptive filter based algorithm
PublicationA technique for improving the quality of speech signals recorded in strong noise is presented. The proposed algorithmemploying adaptive filtration is described and additional possibilities of speech intelligibility improvement arediscussed. Results of the tests are presented.
-
Speech Intelligibility Measurements in Auditorium
PublicationSpeech intelligibility was measured in Auditorium Novum on Technical University of Gdansk (seating capacity 408, volume 3300 m3). Articulation tests were conducted; STI and Early Decay Time EDT coefficients were measured. Negative noise contribution to speech intelligibility was taken into account. Subjective measurements and objective tests reveal high speech intelligibility at most seats in auditorium. Correlation was found between...