Control Speaker Research Articles

Two-stage algorithm of spectral analysis for automatic speech recognition systems

Within the framework of a dynamically developing direction of research in the field of acoustic measurements, the task of spectral analysis of speech signals in automatic speech recognition systems is considered. The low efficiency of the systems in unfavorable speech production conditions (noise, insufficient intelligibility of speech sounds) compared to human perception of oral speech is noted. To improve the efficiency of automatic speech recognition systems, a two-stage algorithm for spectral analysis of speech signals is proposed. The first stage of speech signal processing consists of its parametric spectral analysis using an autoregressive model of the vocal tract of a conditional speaker. The second stage of processing is the transformation (modification) of the obtained spectral estimate according to the principle of frequency-selective amplification of the amplitude of the main formants of the intra-periodic power spectrum. The software implementation of the proposed algorithm based on the high-speed computational procedure of the fast Fourier transform is described. Using the author’s software, a full-scale experiment was carried out: an additive mixture of vowel sounds of the control speaker’s speech with white Gaussian noise was studied. Based on the results of the experiment, it was concluded that the amplitude of the main speech signal formants were amplified by 10–20 dB and, accordingly, a significant improvement in the speech sounds intelligibility. The scope of possible application of the developed algorithm covers automatic speech recognition systems based on speech signal processing in the frequency domain, including the use of artificial neural networks.

Impact of Dysphonic Schoolteachers' Voices on Children's Reaction Times according to Phonemic Contrasts.

This paper aimed at observing the impact of dysphonic voice on children's reception of a linguistic message by evaluating their reaction times (RTs) to instructions given by functional dysphonic and control female schoolteachers (STs). French minimal pairs such as /muʃ/ ("mouche" fly) versus /buʃ/ ("bouche" mouth) embedded in a carrier sentence "click on the drawing of…" were produced by two groups of 10 dysphonic and control female ST, matched in age and year of experience. The phonemical contrasts observed are voicing, nasality, consonantal place of articulation, vowel roundedness, and vowel place of articulation. The experimentation was presented in the form of a computer game to children from 7 to 10 years old. Two images illustrating the target words were presented, accompanied by the oral instructions recorded by ST. With a two-button box created for the experiment, children had to click as quickly as possible on the image corresponding to the instruction. Our results show that the RTs of all children are affected by the ST's dysphonia, regardless of their age and that they have significantly longer RT when discriminating minimal pairs contrasting in voicing when the instruction is given by a dysphonic speaker compared to the same instruction given by a control speaker. These observations could be explained by the fact that functional dysphonia is associated with improper use of the vocal folds and thus an alteration of voicing.

Method for asynchronous analysis of a glottal source based on a two-level autoregressive model of the speech signal

The task of analyzing a glottal source over a short observation interval is considered. The acute problem of insufficient performance of known methods for analyzing a glottal source is pointed out, regardless of the mode of data preparation: synchronous with the main tone of speech sounds or asynchronous. A method for analyzing the glottal source based on a two-level autoregressive model of the speech signal is proposed. Its software implementation based on the high-speed Burg-Levinson computational procedure is described. It does not require synchronization of the sequence of observations used with the main tone of the speech signal and is characterized by a relatively small amount of computational costs. Using the described software implementation, a full-scale experiment was set up and conducted, where the vowel sounds of the control speaker’s speech were used as the object of study. Based on the results of the experiment, the increased performance of the proposed method was confirmed and its requirements for the duration of the speech signal during voice analysis in real time were formulated. It is shown that the optimal duration is in the range from 32 to 128 ms. The results obtained can be used in the development and research of digital speech communication systems, voice control, biometrics, biomedicine and other speech systems where the voice characteristics of the speaker’s speech are of paramount importance.

Circular active noise barrier using theoretical control filter considering interaction between speaker and barrier

A circular active noise barrier using a theoretically calculated control filter without real-time adaptation was proposed for noise reduction in a specific outdoor space. A compact circular barrier is used for the movable system to cope with changes in the location of the workspace, and noise in a wide frequency band can be reduced by conducting active noise control through control speakers arranged around a barrier. However, there was a significant performance gap compared with the maximum performance achieved by using the experimental fixed filter due to an extremely simplified theoretical model which ignores the interaction between the control speakers and the barrier. Therefore, this study tried to minimize the performance degradation when applying the theoretically calculated control filter. Another theoretical model is introduced to improve the noise reduction performance by considering the interaction between the control speaker and the barrier. Through experiment, it is confirmed that noise reduction performance is improved by about 2.6 dB in the frequency of interest.

Public address and sound emission by an active noise control system in ventilation duct networks

In a ventilation duct network, two types of silencers can be used: passive (medium and high frequencies) and active (low frequencies). Active silencers have a loudspeaker (control loudspeaker) that generates an anti-noise. Usually, this system is placed close to the fan and far from the duct air opening. However, when the system is placed close to an air opening, the control speaker can not only be used to reduce the fan noise, but also as a background speaker to play music, information, alerts (the target signal) ... This concept has been implemented in an active noise control (ANC) system based on the adaptive feedforward normalized FxLMS algorithm which uses a reference and an error microphone to pick up respectively the noise to attenuate and the residual noise after control to be minimized. The algorithm is modified so that the target signal does not contaminate the reference nor the error signals. The new algorithm is tested in real time in a short duct with different type of noises and target signals.

Study of the optimal active control of a trombone

In the brass instrument family, the sound can be modified or attenuated using a mute, which is usually inserted in the bell of the instrument. The objective of this paper is to study the principle and the technological feasibility of an active mute using loudspeakers placed in front or around the instrument bell. This mute must reduce the acoustic power emitted by the instrument while avoiding any impact on its playability. At this stage, an optimal control is considered and no real-time controller is implemented. Results show that an active control placed outside the trombone is theoretically feasible and can be efficient to reduce the acoustic power up to 2000 Hz by placing a ring of sources around the bell and a source in front of the trombone. The instrument input impedance is very slightly affected by the control. In accordance with numerical simulations, the experiment showed that placing a control speaker very close in front of the bell of the instrument modifies the pressure field of the instrument in such a way that it allows to obtain a power attenuation greater than the predicted one. The control is technologically achievable but requires a high power for the closest source.

Filter Design for High Frequency Applications

An on chip Radio Frequency(RF) latent band sidestep channel is intended for different groups of frequencies dependent on a three post LC low pass get out and Direct Current (DC) barring capacitors that is prepared by means of Gallium Arsenide (GaAs) Monolithic Microwave consolidated Circuit innovation. The utilization of this model, the RF band skip channel with a tuneable focus recurrence and a legitimate data transfer capacity might be figured it out. A planar winding inductor required on this design and the format of the channel out is limited. The addition misfortune is improved and Substrate misfortunes might be limited. The recreation bring about the schematic and configuration of the band pass channel out that is executed by means of Agilent design software gadget. This sort of LC band pass might be actualized in RF control speaker to the front line report and radar machine for achieving the recurrence determination with an incredible advantage of chosen signal.

Noise reduction performance of active noise control with barrier using theoretical control filter

The external noise is a problem closely related to modern society. There are two ways of noise control methods. One is passive noise control that uses absorbing or insulating materials. The other is active noise control that generates the sound of the opposite phase with the noise by the secondary speakers. Since the effective frequency band of the noise reduction in each method is different, using both methods has been researched to expand the controllable frequency band for dealing with various noise. In order to apply active noise control, the information over the control region is required to obtain a control filter of control speaker. However, the general method that arranges the microphones over the control region can be the obstacles to the user, therefore the other method is needed to obtain the control filter. One way to obtain the control filter without the arrangement of the microphones is using the theoretical model and the control filter can be obtained relatively easily based on the theoretical model. Therefore, the theoretical model is generally used to estimate the maximum performance by simulation. However, there is less research for studying the applicability of the ANC system with a finite noise barrier based on the theoretical model and the verification of the control filter is required by investigating noise reduction performance in the experiment. In this paper, the control filter is obtained based on the theoretical model and noise reduction performance is investigated in the experiment for verification of the control filter.

Fast Active Sound Tuning System for Vehicle Powertrain Response

ABSTRACT This paper describes an active sound tuning (AST) system for vehicle powertrain response. Instead of simply aiming to attenuate cabin interior noise, AST system is capable of reshaping the powertrain response based on predetermined vehicle sound quality criteria. However, conventional AST systems cannot yield a balanced result over the broad frequency range when applied to powertrain noise. It is due to the fact that existing systems are typically configured with the filtered-x least mean square (FXLMS) algorithm or its modified versions, which has inherent frequency dependent convergence behavior due to large dynamic range of secondary path (the electro-acoustic path from the control speaker to the error microphone). Therefore, fast convergence can only be reached at the resonant frequencies. To overcome this inherent limitation, an enhanced adaptive notch filter with inverse model least mean square (ANF-IMLMS) algorithm is proposed as a basis of the AST system for vehicle powertrain response. Compared with the traditional FXLMS algorithm, the proposed algorithm not only increases the convergence speed, but also reduces the computational complexity. Thus, the overall performance of the AST system is improved. Numerical simulation applying measured powertrain data is conducted to validate the performance of the proposed system. Results demonstrate that the proposed system has improved spectral shaping capability with fast convergence.

Yes, You Can? A Speaker’s Potency to Act upon His Words Orchestrates Early Neural Responses to Message-Level Meaning

Evidence is accruing that, in comprehending language, the human brain rapidly integrates a wealth of information sources–including the reader or hearer’s knowledge about the world and even his/her current mood. However, little is known to date about how language processing in the brain is affected by the hearer’s knowledge about the speaker. Here, we investigated the impact of social attributions to the speaker by measuring event-related brain potentials while participants watched videos of three speakers uttering true or false statements pertaining to politics or general knowledge: a top political decision maker (the German Federal Minister of Finance at the time of the experiment), a well-known media personality and an unidentifiable control speaker. False versus true statements engendered an N400 - late positivity response, with the N400 (150–450 ms) constituting the earliest observable response to message-level meaning. Crucially, however, the N400 was modulated by the combination of speaker and message: for false versus true political statements, an N400 effect was only observable for the politician, but not for either of the other two speakers; for false versus true general knowledge statements, an N400 was engendered by all three speakers. We interpret this result as demonstrating that the neurophysiological response to message-level meaning is immediately influenced by the social status of the speaker and whether he/she has the power to bring about the state of affairs described.

Reduction of sound leakage in handheld devices using open loop control

Recently, handheld devices with sound functionality are popular. A problem encountered by using such devices is sound leakage due to inappropriate volume setting, which should be reduced not to disturb people around the user. In previous study, we presented an active control (ANC) technique to control such sound leakage in handheld devices. The least squares (LS) optimum control under various positions of error sensors was investigated. Based on the results, desirable and feasible microphone-loudspeaker setups were suggested. In this paper, we present a novel open loop leakage reduction scheme, and it is compared with adaptive noise control method. To achieve this, control available frequency bands and spatial ranges are studied. And the influence of leakage reduction performance by the receiver and control speaker's radiation pattern are analyzed. Also the effects of the physical environment including the user's hand for the leakage control performance are studied. Finally, the proposed method is implemented on the mobile phone mock-up, and the performance of actual measured leakage reduction is investigated.

능동소음제어를 이용한 고속철도 KTX의 내소음 저감을 위한 모델링에 관한 연구

In this paper, experiments were conducted to validate the importance and necessity of modeling. The modeling was performed using 120㎐ and 280㎐ noise of KTX main interior noise frequency. After the modeling, In order to solve the system instability by the additional path that exists between the control speaker and the error microphone, the secondary path was estimated. Next, simulations were performed to verify the modeling’s necessity and importance. Thought the simulation results, we confirmed that the system with the modeling is more effective for noise reduction than without the modeling.

A Main Concept Analysis for Aphasic Discourse in Irish-English Speakers

Objective: Various methods for the collection of aphasic discourse samples, including single and sequential picture description, storytelling, interactive conversation and monologic recounts, have been reported. The first aim of this study is to examine the cultural appropriateness of the Main Concept Analysis (MCA; Kong, 2009), a discourse analysis developed for Chinese speakers in Hong Kong, for the Irish population. The second aim is to provide preliminary evidence on the clinical applicability of the tool. A case study of using the MCA to assess an Irish-English (IE) speaker with aphasia and a control speaker is presented. Method: Five therapists, who were native IE speakers, were recruited to comment on whether the MCA pictures and scoring criteria were culturally appropriate to the Irish population. The MCA was piloted on six typical native IE speakers and the modified MCA was used to assess one participant with Broca’s aphasia and a matched control to evaluate their performance differences. Main results: The MCA was considered to be culturally appropriate to the Irish population. Preliminary results of the case study indicated that it could differentiate between normal and aphasic discourse production. Conclusion: Future extension should recruit a larger number of participants and should establish reliability and validity of the MCA.

An acoustic study of vowels and coarticulation as a function of utterance type: A case of acquired apraxia of speech

This case study examines vowel production and coarticulation patterns in AOS as a function of utterance type. A female speaker with AOS, and a sex-matched control participant repeated target vowels in three conditions: isolated word, statement frames and question frames. The first two formant frequencies (F1 and F2) of the target vowels were measured at their onset and temporal midpoints. The vowel spaces of both speakers were examined using a two-dimensional vowel chart; temporal midpoint formant frequency values of the first formant (F1) were plotted against the difference between F2 and F1 (F2–F1) as a representation of the constricted versus open (F1) and fronted versus backed (F2–F1) quality of vowels. The speaker with AOS produced vowels that deviated to a posterior/backed distribution within the vowel space. Coarticulation patterns were also examined using F2 locus equations. The speaker with AOS exhibited reduced coarticulation patterns in all three conditions, when compared to the control speaker. Furthermore, these effects were amplified with increasing motoric complexity. The data elucidate the nature of the compromised speech control system in a speaker with AOS.

114 検出マイクロホン・制御スピーカ一体型ANCユニットの開発(騒音・振動の能動制御(3))

Active noise control using the feed-forward method requires an enough long distance (d) between the reference microphone and the control speaker, to satisfy the causality law. In this research, it was proved that the high sampling frequency and the high computation speed could make the distance d very short. A compact active acoustic cell whose d is 10mm was developed by using a flat speaker and a FPGA controller with 100kHz sampling frequency. Even in the case with an ordinary dynamic speaker whose diaphragm is 70mm diameter, the above high speed signal processing system could reduce the distance d down to 60mm. This is effective for realizing the active acoustic shielding system.

하이브리드 제어 알고리즘을 이용한 덕트내 능동소음제어

This study presents the active noise control of duct noise. The duct was excited by a steady-state harmonic and white noise force and the control was performed by one control speaker attached to surface of the duct. An adaptive controller based on filtered x LMS(FXLMS) algorithm was used and controller was defined by minimizing the square of the response of the error microphone. The assemble controller, which is called a hybrid ANC(active noise control) system, was combined with feedforward and feedback controller. The feedforward ANC attenuates primary noise that is correlated with the reference signal, while the feedback ANC cancels the narrowband components of the primary noise that are not observed by the reference sensor. Furthermore, in many ANC applications, the periodic components of noise are the most intense and the feedback ANC system has the effect of reducing the spectral peaks of the primary noise, thus easing the burden of the feedforward ANC filter.

An Articulatory Phonology Perspective on Rhotic Articulation Problems: A Descriptive Case Study

This descriptive case study presents data on a young female speaker (ML) with a history of /r/ sound production problems. Perceptual, acoustic, and kinematic data are provided to illustrate the kind of problems that she is facing in producing this sound, using the Articulatory Phonology model as a theoretical background. Different gestural mechanisms that could explain her kinematic characteristics (and associated acoustic and perceptual features) are evaluated in comparison to published data and new findings from an age-matched control speaker (JE) who performed the same speaking tasks as ML. Results show a well-defined problem in gestural specification for tongue control in /r/ sound productions that can be directly related to changes in acoustics and perception. The analysis demonstrates the feasibility of Articulatory Phonology as a theoretical framework not only for normal speech production, but also to explain potential mechanisms behind changes in articulation in disordered speech.

Air traffic control speech recognition system cross-task and speaker adaptation

We present an overview of the most common techniques used in automatic speech recognition to adapt a general system to a different environment (known as cross-task adaptation) such as in an air traffic control system (ATC). The conditions present in ATC are very specific: very spontaneous, the presence of noise, and high speed speech. So, with a typical speech recognizer the recognition results are unsatisfactory. We have to decide on the best option for the modeling: to develop acoustic models specific to those conditions from scratch using the data available for the new environment, or to carry out cross-task adaptation starting from reliable HMM models (usually requiring less data in the target domain). We begin with a description of the main techniques considered for cross-task adaptation, namely maximum a posteriori (MAP), maximum likelihood linear regression (MLLR), and the two together. We have applied each in two speech recognizers for air traffic control tasks, one for spontaneous speech and the other for a command interface. We show the performance of these techniques and compare them with the development of a new system from scratch. We also show the results obtained for speaker adaptation using a variable amount of adaptation data. The main conclusion is that MLLR can outperform MAP when a large number of transforms is used, and MLLR followed by MAP is the best option. All of these techniques are better than developing a new system from scratch, showing the effectiveness of mean and variance adaptation

347 2次元音場から放射される騒音の能動的制御

Active noise control for a duct system is extended and is applied to that of flat space between two parallel planes. When the noise source has a known periodic, the active noise control can be done by two devices of the control speaker and the error microphone at either duct end, Then filtered-X adaptive notch filer is employed. In this paper, it is tried that the noise radiated from flat panel speakers between two parallel square planes is controlled by the control speaker opposite to the noise source. The flat panel speakers are put at a center part of the board. The control effect in circumference of the square flat space is experimentally verified for two kinds of noise source, sine wave noise and triangular wave noise.

Evidence for separate tonal and segmental tiers in the lexical specification of words: A case study of a brain-damaged Chinese speaker

We present an acoustic study of segmental and prosodic properties of words produced by a female speaker of Chinese with left-hemisphere brain damage. We measured the location of the point vowels /a, e, ə, i, y, o, u/ and determined their separation in the vowel plane, and their perceptual distinctivity. Similarly, the acoustic properties of the four lexical tones were measured in the F0 × time space. The data for our brain-damaged speaker were compared with those of a healthy control speaker. Results show that the patient's vowels hardly suffered from her lesion (relative to the vowel dispersion in the healthy control speaker), but that the identifiability of the four lexical tones was greatly compromised. These findings show that the tonal errors in aphasic speech behave independently of the segmental errors, even though both serve to maintain lexical contrasts in Chinese, and are therefore part of the lexical specification of Chinese words. The present study suggests that the specification of segmental and tonal aspects of lexical entries in Chinese, and in tone languages in general, are located or processed separately in the brain.