Speech Processing Systems Research Articles

Modeling, simulation and implementation of circuit elements in an open-source tool set on the FPAA

An open-source simulator to design and implement circuits and systems, replicating the results from the Field Programmable Analog Array (FPAA) is presented here. The fundamental components like the transistors, amplifiers and floating gate devices have been modeled based on the EKV model with minimal parameters. Systems including continuous-time filters and the analog front-end of a speech processing system have been built from these basic components and the simulation results and the data from the FPAA are shown. The simulated results are in close agreement to the experimental measurements obtained from the same circuits compiled on the FPAA fabricated in a 350 nm process.

Kurtosis Maximization for Blind Speech Separation in Hindi Speech Processing System using Swarm Intelligence and ICA

Blind source separation (BSS) method divides mixed signals blindly without any data on the mixing scheme. This is a main issue in an actual period world whether have to identify a specific person in the crowd or it is a zone of speech signal is removed. Besides, these BSS approaches are collective with shape and also statistical features to authenticate the performance of each one in outline classification. For resolving this issue proposes an active BSS algorithm on the basis of the group search optimisation (GSO). The kurtosis of the signals is used as the objective performance and the GSO is utilised to resolve it in the suggested algorithm. Primarily, source signals are taken into account as the independent component analysis (ICA) to generate the mixing signals to BSS yield the maximum kurtosis. The source signal constituent that is divided out is then smeared off from mixtures with the help of the deflation technique. Each and every source signals establish that important development of the computation amount and the quality of signal separation is attained using the projected BSS-GSO algorithm if associated with the preceding algorithms.

Comparative Analysis of the Mel Frequency Cepstral Coefficients for Voiced and Silent Speech

The extraction of representative characteristics from voice signals, is the basic step for any application where speech signal processing is performed. In this paper, a comparative analysis of the Mel frequency cepstral coefficients, extracted from samples of voiced and silent speech is exposed. The classical methodology for extracting Mel frequency cepstral coefficient is evaluated to show the implementation steps and the robustness of such kind of features representing in a unique form the statistical and representative information of time varying signals. To achieve this comprehensive evaluation, first, a contextualization in theoretical terms of the methodology is performed to show the mathematical basis for the coefficient extraction. Each step of the process is analyzed in terms of the representative information of the coefficients. Therefore an evaluation in comparative terms against the previous developments for data interpretation is done. The coefficients are extracted using the classical implementation of the Mel frequency cepstral coefficient feature extraction process in a free software framework implemented in Matlab®. The results show the robustness of the Mel frequency cepstral coefficient algorithm in the characterization of voiced and silent speech signals, which represents an advantage for the construction of any kind of speech processing systems.

A Review on Feature Extraction Techniques for Speech Processing

Speech and language are considered uniquely human abilities Speech is a complex signal that is characterized by varying distributions of energy in time as well as in frequency, depending on the specific sound that is being produced. The aim of digital speech processing is to take advantage of digital computing techniques to process the speech signal for increased understanding, improved communication, and increased efficiency and Definition of various types of speech classes, feature extraction techniques, speech classifiers and performance evaluation are issues that requires attention in designing of speech processing system.

A technology prototype system for rating therapist empathy from audio recordings in addiction counseling.

Scaling up psychotherapy services such as for addiction counseling is a critical societal need. One challenge is ensuring quality of therapy, due to the heavy cost of manual observational assessment. This work proposes a speech technology-based system to automate the assessment of therapist empathy-a key therapy quality index-from audio recordings of the psychotherapy interactions. We designed a speech processing system that includes voice activity detection and diarization modules, and an automatic speech recognizer plus a speaker role matching module to extract the therapist's language cues. We employed Maximum Entropy models, Maximum Likelihood language models, and a Lattice Rescoring method to characterize high vs. low empathic language. We estimated therapy-session level empathy codes using utterance level evidence obtained from these models. Our experiments showed that the fully automated system achieved a correlation of 0.643 between expert annotated empathy codes and machine-derived estimations, and an accuracy of 81% in classifying high vs. low empathy, in comparison to a 0.721 correlation and 86% accuracy in the oracle setting using manual transcripts. The results show that the system provides useful information that can contribute to automatic quality insurance and therapist training.

Visual influences on the natural referent vowel bias

Research indicates that perceivers (both adult and infant) are universally biased to attend to vowels with extreme articulatory/acoustic properties (peripheral in F1/F2 vowel space). Yet, the nature of this perceptual phenomenon (i.e., the natural referent vowel [NRV] bias) is not fully understood. The present research investigates whether this bias is attributable to general auditory processes or to phonetic processes that track articulatory information available across modalities. In experiment 1, we examined whether adult perceivers are biased to attend to visual information that specifies extreme vocalic articulations. As predicted by the phonetic account, we found a bias favoring relatively more peripheral vowels when only acoustic or only visual speech information was present. In experiment 2, we investigated how the integration of acoustic and visual speech cues influence the effects documented in experiment 1. When acoustic and visual cues were phonetically congruent, a peripheral vowel bias was observed. In contrast, when acoustic and visual cues were phonetically incongruent, this bias was disrupted. Collectively, these results are compatible with the view that the NRV bias is phonetic in nature—the speech processing system appears to be biased toward extreme vocalic gestures, which may be specified in the optic, as well as in the acoustic, signal.

Phonological contribution during visual word recognition in child readers. An intermodal priming study in Grades 3 and 5

This study investigated the phonological contribution during visual word recognition in child readers as a function of general reading expertise (third and fifth grades) and specific word exposure (frequent and less‐frequent words). An intermodal priming in lexical decision task was performed. Auditory primes (identical and unrelated) were used in order to directly activate phonological codes independently of orthographic processing. Overall, the results revealed a widespread phonological priming effect in both grades. There was a significant interaction between grade, priming condition and frequency, revealing that the impact of frequency on identity priming differed between grades. In third grade, the results indicated that the priming effect was greater for less‐frequent than for frequent words. In fifth grade, priming effects were similar for both frequent and less‐frequent words. These findings indicate that print and speech processing systems are interconnected in young readers. Moreover, phonological codes play an important role in word recognition throughout reading development.

Fast Single- and Cross-Show Speaker Diarization Using Binary Key Speaker Modeling

Speaker diarization has become a key process within other speech processing systems which take advantage of single-speaker speech signals. Furthermore, finding recurrent speakers among a set of audio recordings, known as cross-show diarization, is gaining attention in the last years. Current state-of-the-art-systems provide good performance, but usually at the cost of long processing times. This limitation may make current systems not suitable for real-life applications. In this line, the speaker diarization approach based on binary key modeling provides a very fast yet accurate alternative. In this paper, we present the last improvements applied in binary key speaker diarization with the aim of further speeding up the process and improving performance. In addition, we propose a novel method for cross-show speaker diarization based on binary keys. Experimental results show the effectiveness of the proposed improvements for single-show speaker diarization, both in terms of speed and performance, obtaining a real-time factor of 0.0354xRT and a 16.8% relative improvement in performance. Furthermore, our proposed cross-show approach provides very competitive performance, just slightly worse than its single-show diarization counterpart, and exhibits a real time factor of 0.04xRT.

A Fine-Resolution Frequency Estimator in the Odd-DFT Domain

Although many frequency estimation methods are available, few are designed for high-quality speech and audio processing systems, which typically use the modified discrete cosine transform (MDCT) as their analysis filter bank. In this letter, we propose a low complexity frequency estimator that is suitable for MDCT-based systems and that operates in the odd-DFT domain. Taking a complex exponential in noise as the input and deriving the analytical expression of its odd-DFT coefficient, we obtain an interpolated odd-DFT-based frequency estimator. Experiments show that the proposed estimator outperforms all other reported odd-DFT/MDCT domain estimators and has precision that is similar to that of representative DFT domain frequency estimators. The overhead for incorporating this estimator into a speech and audio processing system is small due to the simple odd-DFT to MDCT conversion. The corresponding magnitude and phase estimators are also proposed in this letter.

Universal brain signature of proficient reading: Evidence from four contrasting languages

We propose and test a theoretical perspective in which a universal hallmark of successful literacy acquisition is the convergence of the speech and orthographic processing systems onto a common network of neural structures, regardless of how spoken words are represented orthographically in a writing system. During functional MRI, skilled adult readers of four distinct and highly contrasting languages, Spanish, English, Hebrew, and Chinese, performed an identical semantic categorization task to spoken and written words. Results from three complementary analytic approaches demonstrate limited language variation, with speech-print convergence emerging as a common brain signature of reading proficiency across the wide spectrum of selected languages, whether their writing system is alphabetic or logographic, whether it is opaque or transparent, and regardless of the phonological and morphological structure it represents.

The emergence of phonetic–phonological features in a biologically inspired model of speech processing

Because speech acquisition begins with sensorimotor activity (i.e. babbling and imitation of speech items in order to learn articulatory–acoustic relations) as well as with semantic cognitive processing (i.e. linking phonetic items with concepts), distinctiveness as well as phonetic–phonological features emerge early in speech acquisition. Based on a biologically inspired model of speech processing, using interconnected growing self-organizing maps (I-GSOMs), the phonetic–phonological interface is described here in terms of a numerical computer-implemented model. By simulating early phases of speech acquisition, it can be shown that vocalic features like low–high as well as consonantal features describing the manner of articulation (like plosive, fricative, nasal, lateral, etc.) already arise at sensorimotor levels. This is reflected in our model by an ordering of syllables with respect to phonetic–phonological features at the level of an auditory based neural map. Other features like consonantal place of articulation (labial, apical, dorsal) as well as voiced–voiceless emerge at higher levels within our model, i.e. at the level of neural associations between a phonetic and a semantic neural map. It can be hypothesized from these findings that the phonetic–phonological interface does not appear as a clean cut within the speech processing system but as a broader zone within that system located between sensorimotor and semantic processing.

A Framework for Speech Activity Detection Using Adaptive Auditory Receptive Fields.

One of the hallmarks of sound processing in the brain is the ability of the nervous system to adapt to changing behavioral demands and surrounding soundscapes. It can dynamically shift sensory and cognitive resources to focus on relevant sounds. Neurophysiological studies indicate that this ability is supported by adaptively retuning the shapes of cortical spectro-temporal receptive fields (STRFs) to enhance features of target sounds while suppressing those of task-irrelevant distractors. Because an important component of human communication is the ability of a listener to dynamically track speech in noisy environments, the solution obtained by auditory neurophysiology implies a useful adaptation strategy for speech activity detection (SAD). SAD is an important first step in a number of automated speech processing systems, and performance is often reduced in highly noisy environments. In this paper, we describe how task-driven adaptation is induced in an ensemble of neurophysiological STRFs, and show how speech-adapted STRFs reorient themselves to enhance spectro-temporal modulations of speech while suppressing those associated with a variety of nonspeech sounds. We then show how an adapted ensemble of STRFs can better detect speech in unseen noisy environments compared to an unadapted ensemble and a noise-robust baseline. Finally, we use a stimulus reconstruction task to demonstrate how the adapted STRF ensemble better captures the spectrotemporal modulations of attended speech in clean and noisy conditions. Our results suggest that a biologically plausible adaptation framework can be applied to speech processing systems to dynamically adapt feature representations for improving noise robustness.

Contributions of cerebellar event-based temporal processing and preparatory function to speech perception

The role of the cerebellum in the anatomical and functional architecture of the brain is a matter of ongoing debate. We propose that cerebellar temporal processing contributes to speech perception on a number of accounts: temporally precise cerebellar encoding and rapid transmission of an event-based representation of the temporal structure of the speech signal serves to prepare areas in the cerebral cortex for the subsequent perceptual integration of sensory information. As speech dynamically evolves in time this fundamental preparatory function may extend its scope to the predictive allocation of attention in time and supports the fine-tuning of temporally specific models of the environment. In this framework, an oscillatory account considering a range of frequencies may best serve the linking of the temporal and speech processing systems. Lastly, the concerted action of these processes may not only advance predictive adaptation to basic auditory dynamics but optimize the perceptual integration of speech.

Robust speech recognition system using bidirectional Kalman filter

Kalman filter is normally used to enhance speech quality in a noisy environment, in which the speech signals are usually modelled as autoregressive (AR) process, and represented in the state-space domain. It is a known fact that to identify the changing AR coefficients in every time state requires extensive computation. In this paper, the authors develop a bidirectional Kalman filter and apply it in a speech processing system. The proposed filter uses a system dynamics model that utilises the past and the future measurements to form an estimate of the system's current time state. It provides efficient recursive means to estimate the state of a process that minimises the mean of the squared error. Compared to the conventional Kalman filter, the proposed filter reduces the computation time in two ways: (i) by avoiding the computation of AR parameters in each time state, and (ii) by reducing the dimension of the matrices involved in the difference equations and the measurement equations into constant (1 × 1) matrices. The speech recognition result shows that the developed speech recognition system becomes more robust after the proposed filtering process, and the proposed filter's low computational expense makes it applicable in the practical hidden Markov model-based speech recognition system.

Developmental time course of auditory perception of modulation speech cues

Speech contains strong amplitude modulation (AM) and frequency modulation (FM) cues, which are commonly assumed to play an important role in speech identification for adults. We will review recent studies aiming to characterize the development of auditory perception of AM and FM speech cues for 6 and/or 10-month-old infants learning French or Mandarin [e.g., Cabrera, L., Tsao, F.-M., Gnansia, D., Bertoncini, J., & Lorenzi C. (2014), J. Acoust. Soc. Am. 136, 877–882; Cabrera, L., Bertoncini, J., & Lorenzi, C. (2013), J. Speech, Lang., Hearing Res. 56, 1733–1744]. These studies were based on vocoders, which are analysis and synthesis systems designed to manipulate the modulation components of speech sounds in a given number of frequency bands. Overall, the results suggest that: (i) the auditory processing of AM and FM speech cues is “functional” by 6 months, (ii) the auditory processing of the AM and FM cues is fine-tuned by language exposure between 6 and 10 months. These findings may help improving current models of modulation processing that do not take into account the plasticity of the auditory and speech-processing system.

G2PIL: A Grapheme-To-Phoneme Conversion Tool for the Italian Language

This paper presents a knowledge-based approach for the grapheme to-phoneme conversion (G2P) of isolated words of the Italian language. With more than 7,000 languages in the world, the biggest challenge today is to rapidly port speech processing systems to new languages with low human effort and at reasonable cost. This includes the creation of qualified pronunciation dictionaries. The dictionaries provide the mapping from the orthographic form of a word to its pronunciation, which is useful in both speech synthesis and automatic speech recognition (ASR) systems. For training the acoustic models we need an automatic routine that maps the spelling of training set to a string of phonetic symbols representing the pronunciation.

Three-stage hybrid system for speech signal enhancement

The objective of speech enhancement is to improve the quality and intelligibility of the speech. The goal of this paper is to propose a technique to eliminate the background noise of actual environment in speech processing system. The proposed hybrid speech enhancement method combines both the spectral subtraction algorithm and the conventional Wiener filtering in series with a band pass filter to construct a three–stage hybrid system in spectral domain to enhance the speech with additive noise. The pre–processor is developed for removing noise below hearing threshold. Spectral subtraction and wiener filter removes the background noises. Analysis is done on noisy speech signal database, which consists of speech signals corrupted by eight different real world noises at different SNR levels. The performance of the proposed algorithm is evaluated using different measures such as log–likelihood ratio, signal–to–noise ratio, mean square error and mean opinion score. Experiments results show the increase in the efficiency of the proposed enhancement algorithm.

Feature Extraction Techniques in Speech Processing: A Survey

Speech processing includes the various techniques such as speech coding, speech synthesis, speech recognition and speaker recognition. In the area of digital signal processing, speech processing has versatile applications so it is still an intensive field of research. Speech processing mostly performs two fundamental operations such as Feature Extraction and Classification. The main criterion for the good speech processing system is the selection of feature extraction technique which plays an important role in the system accuracy. This paper intends to focus on the survey of various feature extraction techniques in speech processing such as Fast Fourier Transforms, Linear Predictive Coding, Mel Frequency Cepstral Coefficients, Discrete Wavelet Transforms, Wavelet Packet Transforms, Hybrid Algorithm DWPD and their applications in speech processing.

Native and non-native class discrimination using speech rhythm- and auditory-based cues

In recent years, the use of rhythm-based features in speech processing systems has received growing interest. This approach uses a wide array of rhythm metrics that have been developed to capture speech timing differences between and within languages. However, the reliability of rhythm metrics is being increasingly called into question. In this paper, we propose two modifications to this approach. First, we describe a model that is based on auditory cues that simulate the external, middle and inner parts of the ear. We evaluate this model by performing experiments to discriminate between native and non-native Arabic speech. Data are from the West Point Arabic Speech Corpus; testing is done on standard classifiers based on Gaussian Mixture Models (GMMs), Support Vector Machines (SVMs) and a hybrid GMM/SVM. Results show that the auditory-based model consistently outperforms a traditional rhythm-metric approach that includes both duration- and intensity-based metrics. Second, we propose a framework that combines the rhythm metrics and the auditory-based cues in the context of a Logistic Regression (LR) method that can optimize feature combination. Further results show that the proposed LR-based method improves performance over the standard classifiers in the discrimination between the native and non-native Arabic speech.

A Neural Network based Method for Recognition of Handwritten English Pitman's Shorthand

Pitman Shorthand method of documenting is normally practiced by stenographers to take dictation at speaking speed invented by Sir Isaac Pitman in 1837. Special graphical symbols are used in this method of representing phonetic compositions of the dictated text for certain interval. This shorthand representation itself is a compressed and encrypted format of the English text. The Pitman Shorthand Language (PSL) has the practical advantage of high speed of recording, more than 120–200 words per minute, because of which it is universally acknowledged. This recording medium has its continued existence in spite of considerable developments in speech processing systems, which are not universally established yet. In this method of documenting, speech is directly converted into phonetic strokes, where each phonetic stroke is a composition of consonants and vowels. These consonants are called consonant primitives or simply primitives. In order to exploit the vast transcribing potential of PSL a new area of research on automation of PSL processing is conceived. In this work, the Neural Network Classifier for recognition of PSL at word level is presented. All the activities such as preprocessing of data (image cropping, resizing and implementations) have been carried out using MATLAB. As per this experiment, 94% recognition accuracy is achieved. General Terms Pattern Recognition, Neural Network, Feature Extraction, Classification.