In loudspeaking telephony, one method to prevent howlback is the introduction of voice switching circuits in amplifiers. This method, however, leads to deterioration of simultaneous speech quality. The howlback suppressor presented here, consisting of a pair of complementary comb filters, may be utilized to prevent howlback in loudspeaking telephony without sacrifice of simultaneous speech quality. The comb filter can be derived by digital filter synthesis techniques (Fig. 1). In the experimental system, analogue shift registers, commonly known as the "bucket brigade devices, " are used instead of digital shift registers. The allowable excess loop gain depends on the type of comb response chosen (Table 1). For example, filters with fourth-power cosine and sine amplitude characteristics give 13dB excess loop gain. A modified fourth-power response can give as much as 19dB excess loop gain. This method can also be used for loudspeaking telephony for multi-places. The arrangement of comb filters for three places is shown in Fig. 7. From experiments, it may be said that the speech as heard at the receiving terminal is satisfactory for loudspeaking telephony.
In order to obtain a large acoustic power, it is necessary to drive the diaphragm of the source finitely in accordance with a desired waveform. However, there is not always a linear relation between the motion of the diaphragm and a received signal in the field. In particular, the wave at a receiver distorts as the amplitude at the source is increased. This fact is attributed to the nonlinearity of air, which is neglected for small amplitudes. In this sense, it seems to be important to investigate the nonlinear distortion in the radiation field and, in addition, to derive some simple relation for its phenomenon, if exists. A previous paper indicated the following results for a spherical sound source : the distortion coefficient of sound pressure (the second harmonic sound pressure to the fundamental) is proportional to the velocity of the diaphragm and does not always increase even if sound pressure level is raised. The latter shows the complexity of nearfield distortion. Although some nearfield distortion is usually present, this study shows that farfield distortion, except close to the source, governs the distortion field with increase of the frequency. Then an approximate equation of the distortion coefficient for a source in a free field is derived in neglect of sound absorption. This equation is expressed by some parameters which characterize the sound field. Next, measurements are made in the radiation field of a horn loudspeaker. The results are in satisfactory agreement with the theoretical value.
Problems of nonlinear distortion for the acoustic wave of finite amplitude have been studied for a long time and some of the earlier works dealt with one-dimensional progressive waves in sound absorptionless media. This simplified model, however, does not always give an adequate representation in the majority of the cases of practical interest: analysis including spreading loss and sound absorption is necessary in some cases, for example, in the design of a horn loudspeaker. So this paper presents the theoretical predictions including these effects and experiments for farfield distortion in air. The approximate equation for the distortion coefficient of sound pressure in a free field, which is applicable for waves that do not contain shocks and not for nearfield distortion, is derived under some assumptions by using successive approximation. Then the experimental data on the main axis of a horn loudspeaker are given and compared with the theoretical values in Figs. 2, 3. The results show that the agreement between them is almost satisfactory and falls within the error limits of present work.
A new method is developed for the calibration of acoustic-emission transducers using a reciprocity technique in a Rayleigh-wave sound field. It is applied to practical transducers, and their sensitivity is obtained over a frequency range from 100kHz to 1MHz. The result is compared with the frequency spectrum of an actual acoustic emission signal from a large structure detected by an identical transducer. It is shown that practical characteristics of transducers can be estimated with ease and repeatability in an acoustical environment similar to that of structures such as pressure vessel walls.
In order to evaluate the relationship between hearing loss and effect of residual hearing upon speechreading, and to examine the relationship between informations conveyed visually and auditory, Japanese syllabic articulation scores and confusion matrices were measured under three conditions : visual and auditory reception (alone and combined). As the subjects of this experiment were required to have learned syllables, seven students of a high school for the deaf and six pupils of the upper grade in a hard-of-hearing class in an elementary school were selected, referring to their audiograms (Table 1). Materials used were four sets of 30 syllables consisting of frequently used Japanese syllables and bilabial syllables that serve as cue to speechreading. These syllables were spoken in random order by an adult female and were recorded on videotape. The materials were presented to subjects over a 10 inches black-and-white video monitor placed at 1. 5meters from the subjects under the following three conditions: (1) image only: visual reception, V, (2) speech only: auditory reception, A, and (3) both image and speech: visual-auditory reception, V+A. Syllabic articulation scores obtained under V+A, V, and A conditions are represented (V+A), (V), and (A), respectively, and relations between them and hearing loss are shown in Figs. 1, 2, and 3. (V+A) ranges from 20% to 80%, and decreases with increase of hearing loss. (V) ranges below 40% but shows no apparent relation with hearing loss. On the other hand, (A) ranges below 60% and, similarly to (V+A), decreases with increase of hearing loss. Then, (V+A)-(V) is defined as "effect of residual hearing upon speechreading" and the relationship between this and hearing loss is shown in Fig. 4. It was found in this figure that the effect of residual hearing upon speechreading was inversely related to hearing loss in dB. The effect was more noticable among the pupils attending a hard-of-hearing class. Table 2 shows coufusion matrices among five Japanese vowels obtained under the three conditions: V, A, and V+A. The figure below 4% has been cut off to make it more clear. In case of visual reception, some errors were made near the diagonal. In the case of auditory reception, on the other hand, significant errors were made on the other diagonal. Vowels were recognized fairly well by visual reception, but the percentage of correct response by visual-auditory reception was increased still more. Fig. 5 shows the consonants obtaining the percentage of correct response over 45% under the three conditions: V, A, and V+A. Three phonemes /w, r, m/ were recognized by visual reception and two phonemes /m, k/ by auditory reception. In addition to these four phonemes, 5 more phonemes /p, t, n, h, j/ were recognized by visual-auditory reception. These phonemes were unrecognizable by visual and auditory reception alone but were recognized by combining the two. No other phonemes reached 45% level under any condition. From confusion matrices among vowels and among consonants under the three conditions, the percentage of correct response classified by manner of articulation and place of articulation was calculated and shown in Fig. 6. Visual reception is effective in distinguishing place of articulation and auditory reception is effective in distinguishing manner of articulation, and visual cues and auditory cues complement each other.