Separate Sounds Research Articles

Acoustic quality, proximity, and localization in concert halls: The role of harmonic phase alignment.

Human perception of source distance and direction is predominantly visual. At a music performance, with eyes open, we are usually sure we are hearing the precise location of each musician. But this impression can change dramatically when the eyes are closed. When working on hall acoustics, we find it very useful to walk slowly away from a small group of musicians or actors while looking at the floor. In practice, we often use a small version of the virtual orchestra (Patynen & Lokki, 2011), which plays tirelessly on stage with no visual reference for who is playing which musical line. Close to the group, each performer or musical line can be localized precisely by ear, and the lines, played or spoken, can be easily distinguished separately from the others. As we walk back, very little changes at first. The spread of azimuth diminishes, but the ability to localize and follow the lines is unchanged. The performers are still present.At a particular distance, everything changes. The sound of the ensemble collapses into a fuzzy ball, and the sense of closeness of the performers is lost. We will refer to the distance from the sound sources found in this way as the limit of localization distance, or LLD. The difference in the distance between a position in the hall where the sound is sharp and clear to a position where the sound is fuzzy can be less than a meter or two. One can learn to hear the difference in sound by walking back and forth around the LLD-a new experience for most people. Immediately switching between binaural recordings in front of and behind the LLD can also be startlingly different, even when played through loudspeakers. There is a high degree of agreement between individuals for the distance from the stage of the LLD. The property that creates it appears to be a property of the sound field, not of the individual.It is the sound, the component of a sound field that travels directly from a source to a listener, that contains the information needed to localize the source and to perceive its closeness. We hypothesize that the peaks in the pressure envelope created by the alignment of harmonic phases in the direct sound are essential for the sense of closeness and also enable the ability to separate sounds. We infer from the experience described above that the ability of the ear to separate direct sound from reflections works well down to a certain value of the direct to reverberant ratio, D/R, in a particular venue. Below this value, localization and separation of individual sources become difficult, and the sense of closeness disappears.We are learning that the sense of closeness has a major role for in concert venues. Patynen and Lokki (2011) have developed a laboratory system that can instantly A/B compare different seats and different halls without a moving visual image. They have collected and tested enough hall data to discover a hierarchy of sonic factors that influence preference. Surprisingly, a factor they identified as of great importance to all groups of listeners has been previously unknown. They named the factor proximity, or the impression the listener is sonically close to the performer. To quote from Lokki, Patynen, Kuusinen, and Tervo (2012): An interesting fact is that neither Definition and Reverberance nor early decay time (EDT) and C80 explain the at all. In contrast, the is best explained with subjective Proximity and with Bassiness, Envelopment, and Loudness to some extent. Further there is no objective measure that correlates to Proximity and overall average of preference (p. 3159). In our view, this is a powerful and damning statement. Why do commonly thought-of acoustic qualities such as definition and reverberance, as well as the measures EDT and C80, have no consistent effect on preference, and why has one of the most important determinants of been both previously unknown and unmeasurable?We believe that understanding the acoustics of music venues has been stymied by two problems. …

Source Separation of Heartbeat Sounds for Effective E-Auscultation

This paper proposes a cost effective solution for improving the effectiveness of e-auscultation. Auscultation is the most difficult skill for a doctor, since it can be acquired only through experience. The heart sound mixtures are captured by placing the four numbers of sensors at appropriate auscultation area in the body. These sound mixtures are separated to its relevant components by a statistical method independent component analysis. The separated heartbeat sounds can be further processed or can be stored for future reference. This idea can be used for making a low cost, easy to use portable instrument which will be beneficial to people living in remote areas and are unable to take the advantage of advanced diagnosis methods.

Journal Digest

Journal Digest

Filled Pauses in Modern Irish

In every language there are different types of pauses, the most general division being made between filled and unfilled ones (Trouvain 2003: 27). The latter type either consists of silence or is phonetically filled with breathing, whereas in the former the so-called fillers of different kinds (separate sounds, syllables and even whole words) are used. In Modern Irish filled pauses occur quite frequently in spontaneous speech, especially on the phrase border which can be accounted for by the fact that in such speech the process of formulating thoughts takes place at the very moment of speaking. Thus, it is not uncommon that the verb is chosen as the pause filler due to its initial position. In these cases auxiliary verbs are often used (different forms of the verb ‘to be’ tá, bhí and of the copula is, an). Some other instances include pauses filled by particles (in interrogative sentences) and articles, if a noun takes the initial place in emphatic construction. The present paper presents an analysis of some of such cases based on Connacht Irish material.

Comparative perception of temporally overlapping sounds

Parsing the auditory scene is a problem faced by humans and animals alike. Characteristics such as frequency, intensity, and location all help organisms assign concurrent sounds to specific auditory objects. The timing of sounds is also important for object perception. When sounds totally overlap in time, ascribing separate sounds to individual objects is difficult. However, slight temporal separations make this task slightly easier. In humans, synchronous streams of high and low frequency tones are heard as a single auditory stream. When the tones are slightly offset in time from one another, a second stream emerges. Here, we compared the perception of simultaneous, asynchronous, and partially overlapping streams of tones, human speech sounds, and budgerigar (Melopsittacus undulatus) contact calls in budgerigars and humans using operant conditioning methods. Human and bird subjects identified the partially overlapping stimuli differentially. Both species required less temporal separation to identify the sounds as “asynchronous” for the complex stimuli than for the pure tones. Interestingly, the psychometric functions differed between the two species. These results suggest that both humans and nonhumans are capable of using temporal offsets for assigning auditory objects, and that the ability to do this depends on the spectrotemporal characteristics of the sounds.

Clinical application of speech intelligibility research: The River Street Autism Program at Coltsville.

Introduction Children with autism spectrum disorders (ASD) have severe deficits in communication. Because of this problem, behavioral researchers have sought to develop procedures to improve speech production in children with autism (cf. Koegel & Koegel, 1995; Lovaas, 1977; Lovaas, 2003). These researchers have developed powerful and effective strategies for teaching children to produce initial sounds, words, sentences, and conversational speech. Thus, many children have benefited from these procedures with the development of functional vocal repertoires. While some of these children have adequate intelligibility, there is a portion of this population with compromised intelligibility, due to a limited consonant inventory and less complex syllabic structure (National Research Council, 2001). To address this issue, this article will first review the research in this area and then discuss a model of intervention for children with ASD based on these research findings. Guidelines will be provided on how to: 1) assess intelligibility problems; 2) use the normal developmental model when selecting appropriate targets; 3) sequence treatment objectives; 4) improve intelligibility problems using behavioral interventions; 5) program for generalization; and 6) evaluate program success and troubleshooting. Finally, outcome data for five children on the autism spectrum who have received treatment within the context of this model will be provided. Review of the Research Assessment For children with intelligibility problems, speech language pathologists (SLPs) customarily assess sounds in initial, medial and final positions using picture cues with standardized articulation assessments (Newman & Creaghead, 1989). While this is effective in providing information that will guide treatment for the majority of children with intelligibility issues, these tests pose certain problems for children on the autism spectrum. Specifically, these children do not respond well to traditional assessment tools due to their characteristic language delays, excesses in problem behavior, and depressed motivation. Koegel, Koegel and Smith (1997) reported that modifications to the testing situation to respond to these problems will result in significantly higher test scores. Based on the success of this research, Dyer, Fitzgerald, Harris, and Harris (2005) developed an assessment tool that was designed to respond to these problems by: 1) decreasing demands on expressive language production that are often intrinsic to standardized articulation assessment; 2) providing external motivation with stimulus variation and child preferred items; 3) providing behavioral support through individual behavioral support plans and 4) presenting items in a discrete trial format (cf. Koegel & Schreibman, 1982). In addition, the assessment was delivered by a teacher who had stimulus control in a familiar setting, rather than an SLP who had limited contact with the child. The results of the study showed that this assessment method was more effective in providing information on the child's articulation ability than traditional assessment in the context of reversal designs for four children. This suggests that evaluators take steps to address deficits in motivation and language comprehension in the assessment process in order to gain a more representative sample of children's speech production. Treatment Accumulating studies have documented behavioral intervention techniques in the treatment of articulation for children with autism. Using discrete trial techniques described by Koegel and Schreibman (1982), Dyer, Santarcangelo, and Luce (1987) provided articulation treatment to two adolescents with ASD who had persistent articulation errors. After intervention, the first child met criterion on correct production of three separate sounds in less than 200 trials. The second child met criterion on seven sounds in less than 100 trials. …

音源分離との統合によるミッシングフィーチャマスク自動生成に基づく同時発話音声認識

Our goal is to realize a humanoid robot that has the capabilities of recognizing simultaneous speech. A humanoid robot under real-world environments usually hears a mixture of sounds, and thus three capabilities are essential for robot audition; sound source localization, separation, and recognition of separated sounds. In particular, an interface between sound source separation and speech recognition is important. In this paper, we designed an interface between sound source separation and speech recogniton by applying Missing Feature Theory (MFT) . In this method, spectral sub-bands distorted by sound source separation are detected from input speech as missing features. The detected missing features are masked on recognition not to affect the system badly. Therefore, this method is more flexible when noises change dynamically and drastically. It is the most important issue how distorted spectral sub-bands are detected. To solve the issue, we used speech feature apropriate for MFT-based ASR, and developed automatic missing feature mask generation. As a speech feature, we used a Mel-Scale Log Spectral (MSLS) feature instead of Mel-Frequency Cepstrum Coefficient (MFCC) which is commonly used for ASR. We presented a method of generating missing feature mask automatically by using information from sound source separation. To evaluate our method, we implemented it in a humanoid robotSIG2, and performed the experiments on recognition of three simultaneous isolated words. As a result, our method outperformed conventional ASR with MSLS feature.

Exclusive allocation of a gap that is shared by two separate sounds

When a short temporal gap (40 ms or less) is inserted at the crossing point of two separate sounds of unequal duration, the gap is often allocated only to the shorter sound, whereas the longer sound can be perceptually completed (modal completion). This study shows that the exclusive allocation of the gap to the shorter sound—and the perceptual completion of the longer sound—is most profound when the start of the short sound and the start of the gap are separated by 400 ms or less. A possible explanation for the exclusive allocation of the gap is that within this temporal window, attention is engaged in the formation of a new auditory stream, a process initiated by the shorter sound’s onset. When subsequent auditory elements—even a gap—arrive within the temporal window of attentional engagement, these elements will be automatically included in the new stream. [This study was supported by the COE program of Kyushu University, Japan, and the Japan Society for the Promotion of Science.]

Separating and understanding a talker from a mixture in reverberant spaces

Many studies show that reverberant energy can interfere with the ability to understand a listener by smearing out the temporal modulations and distorting the spectral content that convey speech meaning. However, the amount of distortion caused by the reverberation in everyday settings often is not severe enough to degrade speech understanding when there is only one talker in a room. Despite this, even modest reverberation can have an enormous impact on the ability to understand the same talker when there is another sound source in the environment. One factor that likely contributes to the problem of understanding a talker in a mixture of sounds in reverberant spaces is that it is difficult to separate the target talker from the competing sounds. This talk will explore the ways in which reverberant energy degrades the ability to separate sounds in a mixture. The effects of reverberation on different cues for source segregation, including pitch and location, will be discussed. [Work supported by AFOSR and NIH.]

Performance of the Vaisala RS80A/H and RS90 Humicap Sensors and the Meteolabor “Snow White” Chilled-Mirror Hygrometer in Paramaribo, Suriname

Abstract In climate research there is a strong need for accurate observations of water vapor in the upper atmosphere. Radiosoundings provide relative humidity profiles but the accuracy of many routine instruments is notoriously inadequate in the cold upper troposphere. In this study results from a soundings program executed in Paramaribo, Suriname (5.8°N, 55.2°W), are presented. The aim of this program was to compare the performance of different humidity sensors in the upper troposphere in the Tropics and to test different bias corrections suggested in the literature. The payload of each sounding consisted of a chilled-mirror “Snow White” sensor from Meteolabor AG, which was used as a reference, and two additional sensors from Vaisala, that is, either the RS80A, the RS80H, or the RS90. In total 37 separate soundings were made. For the RS80A a clear, dry bias of between −4% and −8% RH is found in the lower troposphere compared to the Snow White observation, confirming the findings in previous studies. A mean dry bias was found in the upper troposphere, which could be effectively corrected. The RS80H sensor shows a significant wet bias of 2%–5% in RH in the middle and upper troposphere, which has not been reported before. Comparing observations with RS80H sensors of different ages gives no indication of sensor aging or sensor contamination. It is therefore concluded that the plastic cover introduced by Vaisala to avoid sensor contamination is effective. Finally, the RS90 sensor yields a small but significant wet bias of 2%–3% below 7-km altitude. The time-lag error correction from Miloshevich et al. was applied to the Vaisala data, which resulted in an increased variability in the relative humidity profile above 9- (RS80A), 8- (RS80H), and 11-km (RS90) altitude, respectively, which is in better agreement with the Snow White data. The averaged Snow White profile is compared with the average profiles of relative humidity from the European Centre for Medium-Range Weather Forecasts (ECMWF). No significant bias is found in either the analyses or the forecasts. The correlation coefficient for the Snow White and ECMWF data between 200 and 800 hPa was 0.66 for the 36-h forecast and 0.77 for the analysis.

Auditory Scene Analysis and the Role of Phenomenology in Experimental Psychology.

Abstract This paper is about two topics: a) the role of subjectivity in psychological research; and, b) my research on the perceptual organization of sound, in which subjectivity has played an important role. Audio demonstrations that appeal to the subjective experience of the reader are presented in lieu of objective research data to support the claims made about auditory organization. It is argued that all psychological research depends on an underlying framework of intuition (unformalized knowledge acquired through everyday experience), and that this intuition plays a role in the design of the experiment. Personal Experience as Data The personal experience of the researcher has not fared well in scientific psychology. Since the failure of Titchener's Introspectionism in the early 20th century, and the rise of Behaviourism, scientific psychology has harboured a deep suspicion of the experience of the researcher as an acceptable tool in research. One would think that the study of perception would be exempt from this suspicion, since the subject matter of the psychology of perception is supposed to be about how a person's experience is derived from sensory input. Instead, academic psychology, in its behaviouristic zeal, redefined perception as the ability to respond differently to different stimuli - bringing it into the stimulus-response framework. Despite Behaviourism's fall from grace, psychology still insists on a behaviouristic research methodology. In my own research, however, subjectivity has played a central role. It was a perceptual experience that got me going on the topic of perceptual organization in the first place: I was preparing an experiment on learning, involving a rapid sequence of unrelated sounds, each about the length of a speech phoneme. I spliced together one-tenth-second segments of many different sounds - water splashing in a sink, a dentist's drill, a tone, a vowel, etc. When I played the tape back to myself, though, I did not experience the sequences in the order that they were recorded on the tape. It appeared that nonadjacent sounds were grouping together and appeared to be adjacent. It was the similar sounds that seemed to be forming integrated perceptual sequences. This reminded me of an essay I had written at the University of Toronto on the topic of Gestalt Psychology. Some of the Gestaltist examples showed that similar visual forms would group together and segregate from dissimilar ones. Perhaps an analogous sort of grouping might be happening in my auditory sequence. Although I had never been trained in auditory perception research, this one subjective experience set me off on a 36-year period of study. When I use the term phenomenology, it is not with the technical meaning it has in the writings of Husserl or Heidegger, but is just a fancy name for experience. In all my years of research on the conditions under which a mixture of sounds will blend or be heard as separate sounds, my own phenomenology has played a central role in deciding what to study and how to study it. Also, the subjective experiences of colleagues and students have made it possible for them to understand the phenomena by listening to auditory demonstrations. I almost never carried out a study whose outcome I did not know in advance by listening to the stimuli. Only when I had figured out the conditions that would give rise to the effect I wanted to study, would I design a formal experiment. It is impossible to overestimate how many years of research this saved. It made it possible to study a large set of theoretical issues in perceptual organization without elaborate sets of preliminary experiments to establish the right parameters. The approach of listening to many variants of the signals and getting familiar with their effects at a personal level permitted us to speed up the development of a general overview of auditory grouping (Bregman, 1990/1994), rather than to merely accumulate more and more highly quantitative knowledge about a narrow experimental paradigm - a not-unknown practice in experimental psychology. …

Improvement of recognition of simultaneous speech signals using AV integration and scattering theory for humanoid robots

This paper presents a method to improve recognition of three simultaneous speech signals by a humanoid robot equipped with a pair of microphones. In such situations, sound separation and automatic speech recognition (ASR) of the separated speech signal are difficult, because the signal-to-noise ratio is quite low (around −3 dB) and noise is not stable due to interfering voices. To improve recognition of three simultaneous speech signals, two key ideas are introduced. One is two-layered audio–visual integration of both name (ID) and location, that is, speech and face recognition, and speech and face localization. The other is acoustical modeling of the humanoid head by scattering theory. Sound sources are separated in real-time by an active direction-pass filter (ADPF), which extracts sounds from a specified direction by using the interaural phase/intensity difference estimated by scattering theory. Since features of separated sounds vary according to the sound direction, multiple direction- and speaker-dependent acoustic models are used. The system integrates ASR results by using the sound direction and speaker information provided by face recognition as well as confidence measure of ASR results to select the best one. The resulting system shows an improvement of about 10% on average against recognition of three simultaneous speech signals, where three speakers were located around the humanoid on a 1 m radius half circle, one of them being in front of him (angle 0°) and the other two being at symmetrical positions (± θ) varying by 10° steps from 0° to 90°.

Analysis and modelling of overtone singing in the sygyt style

Overtone singing, also called biphonic singing, xöömij or chant diphonique in french; is a special singing style that exhibits two or more separate sounds – one “drone” sound of relatively low pitch and one or more high pitch melody sounds. The perceived pitches of the upper tones are multiples of the drone sound, i.e. taken from its overtone scale. Compared to voiced sounds of western style singers, the relative amplitude of the melody pitches is quite high, and the formant bandwidth of overtone sounds is small. This paper tries to answer the question of how these formant properties are achieved. Experimental investigations and numerical calculations prove the existence of two closely neighboured formants for the production of the melody sound in the sygyt style.

Responses of auditory nerve fibers to harmonic and mistuned complex tones

Responses of auditory nerve fibers were obtained to harmonic complex tones in which single components could be mistuned. Human listeners hear the harmonic tones as single sounds, but the same tones with one component mistuned are heard as two separate sounds. Fourier analysis of the temporal discharge patterns indicated that auditory nerve fibers typically responded to one or two stimulus components near the fibers’ characteristic frequencies. At low stimulus levels, the discharge patterns could also exhibit low-frequency modulation that was produced by beating of two higher-frequency components. The same components were observed in the response spectra, whether those components were part of the original harmonic series or had been mistuned. The discharge patterns and response spectra were consistent with expectations based on previous studies of auditory nerve fibers with harmonic tones and other complex sounds. However, the discharge patterns differed dramatically from the discharge patterns elicited from inferior colliculus neurons by comparable stimuli [Sinex et al., Hear. Res. 168 (2002) 150–162].

Responses of inferior colliculus neurons to harmonic and mistuned complex tones

Responses of inferior colliculus neurons to simplified stimuli that may engage mechanisms that contribute to auditory scene analysis were obtained. The stimuli were harmonic complex tones, which are heard by human listeners as single sounds, and the same tones with one component ‘mistuned’, which are heard as two separate sounds. The temporal discharge pattern elicited by a harmonic complex tone usually resembled the same neuron’s response to a pure tone. In contrast, tones with a mistuned component elicited responses with distinctive, stereotypical temporal patterns that were not obviously related to the stimulus waveform. For a particular stimulus configuration, the discharge pattern was similar across neurons with different pure-tone frequency selectivity. A computational model that compared response envelopes across multiple narrow bands successfully reproduced the stereotypical response patterns elicited by different stimulus configurations. The results suggest that mistuning created a temporally synchronous distributed representation of the mistuned component that could be identified by higher auditory centers in the presence of the ongoing response produced by the remaining components; this kind of representation might facilitate the identification of individual sound sources in complex acoustic environments.

Separation of harmonic sounds using multipitch analysis and linear models for the overtone series

A signal processing method for the separation of concurrent harmonic sounds is described. The method is based on a two-stage approach. First, a multiple fundamental frequency estimator is applied to find initial sound parameters which are reliable, but inaccurate and static. Second, time-varying sinusoidal parameters are estimated in an iterative algorithm. The harmonic structure is retained by keeping the frequency ratio of overtones constant over time. Overlapping harmonic components are resolved using linear models for the overtone series. In practice, the models retain the spectral envelope continuity of natural sounds. Simulation experiments were carried out using generated test signals, which were random mixtures of two to six notes from recorded natural instruments. The system is able to produce meaningful results in all polyphonies, the quality of separated sounds gradually degrading along with the polyphony. Some denoising algorithms were applied to suppress nonstationary noise component, such as drums in real-world music signals. However, the usability of the system for real musical signals is still quite limited.

Method for detecting, recording and deterring the tapping and excavating activities of woodpeckers

An apparatus for deterring woodpeckers from tapping on and excavating within a utility pole or the like includes a housing having mounting flanges for direct attachment to the pole so as to receive vibrations therefrom. A transducer is attached to the mounting wall so as to convert the vibrations preferentially over airborne sounds into signals. A circuit compares the vibrations with a long term average and emits an output in response to detection above a threshold, whereupon the outputs are counted and if the number within a predetermined time exceeds a preset minimum a sound transmitter is actuated for emitting a deterrent sound externally of the housing. A memory contains a plurality of separate audible deterrent sounds including some of the same species and some of predators which act as a deterrent to the woodpeckers and these are selected sequentially and transmitted after a delay to avoid habituation. The power source includes a solar cell charging a battery and to maintain battery power the circuit is arranged to partially power down between detecting periods, to prevent sound actuation at certain times of day and certain times of year and to issue deterrent sounds only once for every set time period.

Geophysical investigations of the Lake Lappajärvi impact structure, western Finland

A circular negative Bouguer anomaly approximately −10 mgal in amplitude and 17 km in diameter is associated with the Lappajärvi impact structure. The related mass deficit is −440 × 10 11 kg. There are irregularities in the anomaly pattern due to differentially faulted bedrock blocks and lateral density contrasts between bedrock, shocked bedrock, breccia, suevite, impact melt rock and overburden. The minimum amount of impact melt rock was interpreted, from the positive residual Bouguer anomaly between the centre and the rim in the northwestern part of the crater, as being 1.5–2.0 km 3. There are a few small aeromagnetic anomalies (up to 200 nT at an altitude of 40 m) within the crater, which is mostly characterized by a lack of magnetized material. Strong magnetic anomalies (a few thousand nT) which are abundant in the surroundings disappear in the crater area. The aero-electromagnetic (AEM) data show strong anomalies with inphase /quadrature ratios close to 1, indicating moderately conducting material within the crater. An electromagnetic wide-band (SAMPO EM) profile was measured across the annular depression between the centre and the rim of the crater. Three layers were interpreted. The thickness of the uppermost layer, mostly conductive overburden, is about 100–200 m and the resistivity 60–170 Ωm. In the annular depression, beneath the uppermost layer, there is a 50–250 m thick layer of suevites and breccias with resistivity of about 17–50 Ωm. The resistive lowest layer follows the base of breccia and suevite. Some separate soundings were carried out near the centre of the crater. According to one of these soundings, there is a conductive layer of suevite and breccia, with a resistivity of 30–40 Ωm, beginning at depth of 150 m beneath the exposed impact melt rock. The results of geophysical interpretations, such as the annular depression between the centre and the rim of the crater, the small thickness/diameter ratio of the geophysically anomalous part of the crater, and the calculated amount of impact melt rock, indicate that the Lappajärvi structure belongs to the category of complex craters. Rhomboidal geophysical anomaly patterns (in particular AEM anomalies), parallel to old bedrock structures suggest that the collapse of the transient crater and/or subsequent crater modification were controlled by NW-SE and NE-SW trending pre-existing fracture and fault zones.

Magnetotelluric measurements across the eastern neovolcanic zone in south Iceland

A magnetotelluric survey consisting of 18 separate soundings was performed in 1982 along a 125‐km profile crossing the eastern neovolcanic zone in south Iceland. Although the three‐dimensional ocean effect may be significant at the longest periods which previous workers have analyzed to study features in the upper mantle beneath Iceland, our studies suggest that this effect is minimal for periods less than 1000 s at distances greater than 50 km from the coast. Therefore we are confident that two‐dimensional interpretations are valid at periods suitable for interpreting crustal resistivities. Such a two‐dimensional interpretation presented here indicates the presence of a layer (∼5 km thick) of low resistivity (5 ohm m) at the crust‐mantle interface (10–20 km depth). The depth to this layer increases from the center of the neovolcanic zone. This feature has been seen elsewhere in Iceland by other workers. However, the present study suggests that east of the neovolcanic zone the low‐resistivity layer is not present, which we argue may be associated with a recent eastward shift of the axial rift zone to a new location within older crust. A striking feature of our two‐dimensional crustal interpretation is that pockets of low‐resistivity material appear to be present at approximately 8 km depth at the base of the crust beneath the two volcanic centers that we have studied, Torfajokull and Hekla. These may be manifestations of local magma chambers.

Speech sound distortions caused by changes in complete denture morphology.

A numerical method for speech analysis has been developed. The method is based on sounds spectrograms as carriers of information about individual voice characteristics. The effects on speech sounds caused by alterations of the oral cavity dimensions with complete dentures of different morphology were analysed. Different positions of upper incisors and different thickness of the denture palatal plate were used as changeable factors in the longitudinal and transverse dimensions of the oral cavity. The effects of these changes on the relative duration of separate sounds in a word and the patient adaptation period were also analysed. From the results some conclusions of pragmatic value were obtained useful in the clinical procedures of denture construction.