Bradford N. Gover

Measurements of directional properties of reverberant sound fields in rooms using a spherical microphone array

The directional variation of sound at a point has been studied in three rooms, using a measurement system described previously [J. Acoust. Soc. Am. 112, 1980–1991 (2002)]. The system uses a pair of 32-element spherical microphone arrays to obtain directional impulse responses in each of 60 steering directions, with an angular resolution of 28°, covering all directions in the whole solid angle. Together, the array measurements span the frequency range from 300 to 3300 Hz. The angular distribution of incident sound energy is visualized on a three-dimensional plot, and quantified by computing the directional diffusion and the directional peak-to-average level difference (“anisotropy index”) of the sound field. The small-to-medium-sized rooms had reverberation times of 360, 400, and 600 ms. Measurements were made for several source and receiver locations in each, and were analyzed over several time ranges (full decay time of room, late time decay, 2-ms windows throughout the decay). All measured sound fields ...

Microphone array measurement system for analysis of directional and spatial variations of sound fields

A measurement system has been developed that is capable of analyzing the directional and spatial variations in a reverberant sound field. A spherical, 32-element array of microphones is used to generate a narrow beam that is steered in 60 directions. Using an omnidirectional loudspeaker as excitation, the sound pressure arriving from each steering direction is measured as a function of time, in the form of pressure impulse responses. By subsequent analysis of these responses, the variation of arriving energy with direction is studied. The directional diffusion and directivity index of the arriving sound can be computed, as can the energy decay rate in each direction. An analysis of the 32 microphone responses themselves allows computation of the point-to-point variation of reverberation time and of sound pressure level, as well as the spatial cross-correlation coefficient, over the extent of the array. The system has been validated in simple sound fields in an anechoic chamber and in a reverberation chamber. The system characterizes these sound fields as expected, both quantitatively from the measures and qualitatively from plots of the arriving energy versus direction. It is anticipated that the system will be of value in evaluating the directional distribution of arriving energy and the degree and diffuseness of sound fields in rooms.

Measures for assessing architectural speech security (privacy) of closed offices and meeting rooms

Objective measures were investigated as predictors of the speech security of closed offices and rooms. A new signal-to-noise type measure is shown to be a superior indicator for security than existing measures such as the Articulation Index, the Speech Intelligibility Index, the ratio of the loudness of speech to that of noise, and the A-weighted level difference of speech and noise. This new measure is a weighted sum of clipped one-third-octave-band signal-to-noise ratios; various weightings and clipping levels are explored. Listening tests had 19 subjects rate the audibility and intelligibility of 500 English sentences, filtered to simulate transmission through various wall constructions, and presented along with background noise. The results of the tests indicate that the new measure is highly correlated with sentence intelligibility scores and also with three security thresholds: the threshold of intelligibility (below which speech is unintelligible), the threshold of cadence (below which the cadence of speech is inaudible), and the threshold of audibility (below which speech is inaudible). The ratio of the loudness of speech to that of noise, and simple A-weighted level differences are both shown to be well correlated with these latter two thresholds (cadence and audibility), but not well correlated with intelligibility.

Some spatial and temporal effects on the speech privacy of meeting rooms.

This paper reports on initial experiments concerning how key spatial and temporal effects in rooms influence the speech privacy provided by enclosed rooms. The first part of the work demonstrates that for the same signal-to-noise ratio, the intelligibility of speech and the threshold of intelligibility are significantly different for transmission between real rooms than in the previous results in approximately free-field conditions [B. N. Gover and J. S. Bradley, J. Acoust. Soc. Am. 116, 3480-3490 (2004)]. The second part investigates the influence of aspects of the spatial and temporal components of sound fields in typical rooms, to explain these differences for transmission between real rooms. These components included the separate effects of early-arriving and later-arriving reflected speech sounds. They also included the effects of spatially separated speech and noise sources as well as more diffuse noise representative of typical meeting rooms. In realistic combinations these effects are of practical importance and can change privacy criteria by 5 dB or more. Ignoring them could lead to costly over-design of the sound insulation required to achieve adequate speech privacy.

An artificial neural network approach for predicting architectural speech security (L)

Signal-to-noise type measures have been developed for predicting architectural speech privacy and speech security, which is required to accurately rate the probability of a listener outside a room being able to overhear conversations from within the room. However, these measures may not be ideal for speech security situations. In the present work, an approach that uses the artificial neural networks to directly represent the functional relationship between the octave band (250 Hz–8 kHz) S/N ratios and the speech intelligibility score and security thresholds has been investigated. The artificial neural network approach provides a direct and accurate method for predicting the speech intelligibility score and security thresholds.

Comparison of speech intelligibility scores for direct listening and headphone playback

Speech recognition test scores from direct listening and headphone playback were compared to validate the use of headphone playback for investigating the effects of varied room acoustics conditions. Varied acoustical conditions for direct listening, that were representative of classrooms, were achieved using simulated sound fields in an anechoic room with an 8‐channel electro‐acoustic simulation system. Eight conditions consisting of the combinations of 2 reverberation times and 4 signal‐to‐noise ratios were used. By recording these conditions with an acoustical mannequin, headphone playback of the same conditions could be achieved and the corresponding speech recognition scores compared. Two different approaches for correcting for the microphone and headphone responses were compared. Subjects also responded to a short questionnaire to describe the audible differences between direct playback and headphone listening. Because spatial characteristics were frequently found to differ between the two listening ...

Assessment of Architectural speech security of closed offices and meeting rooms

The degree to which converstions taking place within closed offices and meeting rooms are audible or intelligible to a listener outside the room depends on the listeners hearing ability, the talkers voice level, the sound attenuation afforded by the building, and the background noise at the position of the listener. Sensible assumptions can be made about the listeners ability so that signal‐to‐noise measures can be used to reliably indicate audibility or intelligibility. A survey of actual meetings was conducted to acquire estimates of typical voice and noise levels, which can be used in the prediction of security. In existing buildings, the background noise can be measured, leaving only the need to measure sound attenuation through the building. A new measurement prodcedure has been developed that accurately indicates the attenuation from within a source room to a single receiving point outside the room. The results of this procedure can be used to derive a reliable indicator of the level of speech secur...

Measurement of directional information in sound fields

The directional characteristics of sound arriving at a listening point in a room are being studied with a view to improving microphone pickup of speech in teleconference rooms. Knowledge of the directions from which the direct, reflected, and reverberant sounds arrive is expected to be useful in microphone array design. At present, there exist several techniques which attempt to measure the direction of sound arrival, based on comparisons and postprocessing of multiple microphone signals. This postprocessing can involve intensity analysis, deconvolution analysis, or cross‐correlation analysis. Each of these approaches has its benefits, drawbacks, and tradeoffs, and may perform better or worse in a given situation. Studies of several of these directional measurement techniques are underway to ascertain their usefulness and validity in a variety of situations, through both simulations and real room measurements.

Setting speech privacy criteria for closed rooms in terms of speech privacy class values.

The new ASTM E2638 standard defines speech privacy class (SPC) as the sum of the measured average noise level at the position of a potential eavesdropper outside the room and the measured level difference between a source room average and the transmitted levels at the same potential eavesdropper location. For a given situation, the likelihood of transmitted speech being audible or intelligible can be related to the probability of higher‐speech levels occurring in the meeting room. Increasing speech privacy criteria can be defined in terms of increasing SPC values. For a particular meeting room speech level, there is an SPC value for which transmitted speech would be below the threshold of intelligibility or even below the threshold of audibility. One can therefore create a set of increasing SPC values corresponding to increasing speech privacy and for each SPC value, one can give the probability of transmitted speech being either audible or intelligible. This paper describes a new procedure that makes it ...

Measurement of speech privacy of closed rooms using the American Society for Testing and Materials ASTM E2638.

The degree of speech privacy between various locations in buildings has been of increasing concern, driven in part by legislative requirements and the desire for increased security. Until recently, ASTM E1130 “Standard Test Method for Objective Measurement of Speech Privacy in Open Plan Spaces Using Articulation Index” has been the only standardized measurement method available. This test method is, for a variety of reasons, not suitable for assessing speech privacy of closed rooms or conditions of very high‐speech privacy. These issues are, however, addressed by the new ASTM E2638 “Standard Test Method for Objective Measurement of the Speech Privacy Provided by a Closed Room.” ASTM E2638 describes a test method suitable for enclosed rooms of nearly all sizes and defines a measure called speech privacy class that can be used to accurately rate speech privacy over a very broad range, from no privacy to very high secrecy. The new E2638 measurement method will be described, and measurements in several real rooms will be presented.