Word Intelligibility Tests Research Articles

Methods for improving word intelligibility of bone-conducted speech by using bone-conduction headphones

Bone-conduction (BC) headphones enable listeners to hear sounds through BC while leaving the ear canal (EC) open to pass surrounding air-conducted (AC) sound at the same time. However, BC transmission degrades the word intelligibility of BC speech, especially in noisy environments. This article examines the attenuation of higher-frequency components during BC transmission as a key characteristic that degrades speech intelligibility and proposes two methods for improving the intelligibility of BC speech by means of BC headphones. We focused on two types of BC transfer function, namely vibration of the regio temporalis (RT) and sound radiated in the EC relative to AC speech sound, and developed two methods for each (first order high-frequency emphasis (FOE) and higher-order high-frequency emphasis (HOE)) for improving the intelligibility of BC speech by compensating for the BC transmission characteristics. Speech signals in the proposed methods are filtered using four types of emphasis filter (RT-FOE, EC-FOE, RT-HOE, and EC-HOE). The results of word-intelligibility tests demonstrated that the word recognition rate, especially with FOE by the inverse transfer function between RT vibration and AC speech (RT-FOE), tended to significantly increase compared to that with no emphasis. Our findings suggested that linear compensation for the attenuation of higher frequency due to BC transmission by using the inverse transfer function between RT vibration and AC speech, is effective for improving BC speech intelligibility.

Intelligibility of synthetic words generated by transformation of a sequence of discrete acoustic events into modulation of the vocal tract shape

Within the paradigm of a recent model of speech production [Story and Bunton, JASA 146(4), 2522–2528], an utterance is specified as a sequence of relative acoustic events along a time axis. These events consist of directional changes of the vocal tract resonance frequencies called resonance deflection patterns (RDPs) that, when associated with a temporal event function, are transformed via acoustic sensitivity functions, into time-varying modulations of the vocal tract shape. RDPs specifying the targeted directional shift of the first three resonances for bilabial, alveolar, and velar consonants would be coded as [−1 −1 −1], [−1 1 1], and [−1 1 −1], respectively. In this study, these RDPs were combined with four vowels (“ih, ae, eh, uh”) to construct a set of 40 American English words (CVCs). A word intelligibility test was conducted in which listeners heard a synthesized target word and were asked to indicate what they heard by choosing a word from a matrix that included the target and seven near-neighbor words. Results indicate listener word recognition was aligned with the RDP settings, suggesting that they are an effective discrete representation of phonetic segments that can be transformed into speech by modulation of the vocal tract shape.

Characteristics of speech perception by distantly-presented bone-conducted ultrasound assessed by word intelligibility and monosyllable articulation tests.

High-frequency sounds above 20 kHz presented via bone conduction can be heard clearly and transmit speech information using amplitude modulation. Additionally, bone-conducted ultrasound (BCU) can be perceived even when the vibrator is presented to body parts distant from the head, such as the neck, arm, and trunk. To evaluate this previously presented BCU hearing, word intelligibility and monosyllable articulation tests were conducted in Japanese. The results suggested that a practical speech transmission, comparable to ordinary BCUs presented onto the head, can be obtained by distantly presented BCU.

Assessment of the difference limens for frequency, monosyllable articulation and word intelligibility by distantly-presented bone-conducted ultrasound.

Ultrasound can be clearly perceived by bone-conduction, and this "bone-conducted ultrasound (BCU)" can transmit speech information by using amplitude modulation (AM). Further, BCU can be perceived not only on the head but also on the distal parts of the body like the neck, trunk and arms. This "distantly-presented BCU" can be applied to the novel interface that can transmit sound information selectively to specific users who touches the vibrator. However, the ability to transmit sound information of distantly-presented BCU is unclear. First, to assess frequency discrimination ability, difference limens for frequency (DLFs) of the distantly-presented AM-BCU were measured with/without a low-pass masking noise that masked the self-demodulated components generated by the nonlinearity of biological tissues. DLFs comparable to that of air-conducted sounds were observed, whereas DLFs significantly increased above 1 kHz under the masking condition. These results suggest that practical frequency discrimination ability can be obtained even when BCUs were presented to distal body parts. Additionally, it is indicated that the demodulated components may contribute to transmitting frequency information above 1 kHz. Second, monosyllable articulation and word intelligibility tests were conducted in Japanese. The intelligibility and articulation at the neck were 55% and 38% respectively, whereas they decreased as the stimulus placement gets farther from the head. The results suggest the distantly-presented BCU device can be applied to transmission of speech information.

3122 Longitudinal Recovery of Speech Motor Function Following Facial Transplantation

OBJECTIVES/SPECIFIC AIMS: Using a novel biomechanical-based motor speech assessment alongside commonly used clinically-based motor speech assessments, the goal of this study was to describe longitudinal recovery in speech movements and functional speech in a cohort of 5 patients following facial transplantation. METHODS/STUDY POPULATION: Five participants who had received either full or partial face transplantation were included in this study. Each participant received a unique facial graft from their donor, which included varied amounts of soft tissue, facial musculature, nerve, and bone. Two participants were early in the recovery period and were assessed from zero to 24 months post-transplantation. Three participants were late in the recovery period and were assessed from 36 to 60 months post-transplantation. Each participant completed two data collection sessions and the average time between sessions was 20.4 months. At each session, orofacial movements were recorded using a 3D motion capture system. A 4-sensor head marker was used to subtract head movement (translation and rotation) from the facial markers. The analyses in this study were restricted to two markers: midline lower lip and a virtually calculated midline jaw marker. A marker at the top of the nose bridge was used as the origin point. The following kinematic variables were obtained from each lip-jaw movement time-series: peak movement speed (mm/s), and displacement (mm). Each patient was instructed to perform 10 repetitions of the phrase “buy bobby a puppy” at his or her typical speaking rate and volume. Sentence-level intelligibility was obtained using the Sentence Intelligibility Test (SIT) and word-level intelligibility was obtained using the Word Intelligibility Test, using standard procedures. Intelligibility, measured in percentage of words correctly transcribed, and speaking rate, measured in words per minute (wpm), was derived from the SIT sentences for each patient. Intelligibility, measured in percentage of words correctly chosen via multiple choice was derived from the Word Intelligibility Test. RESULTS/ANTICIPATED RESULTS: Effect sizes (Cohen’s d) across the 10 trials of “buy bobby a puppy” were computed to assess the effects of recovery time on range of motion and speed of the lower lip alone, the jaw alone, and the lower lip and jaw together for both range of motion and for speed. The largest effect sizes were observed for increased range of motion and increased speed of the articulators for participants within 24 months of surgery. Smaller effect sizes were observed for these parameters for the participants in the later stages of recovery, with some participants showing declines in range of motion and speed of some but not all articulators. Descriptive statistics indicate that both speech and word intelligibility improvements are most notable in the first two years following transplantation and appear to plateau during the later stages of recovery. Only two out of five of our participants achieved “normal” speech intelligibility (i.e., >97%) at five years post-transplantation. DISCUSSION/SIGNIFICANCE OF IMPACT: Biomechanical assessment revealed that kinematic recovery of articulator range of motion and speed appears most significant in the first two years following surgery, but that improvement continues to some degree as far as five-years post-transplant. Clinically-based assessments suggest that gains in intelligibility appear to plateau by 3-years post-surgery.

Effects of inter-word pauses on speech intelligibility under long-path echo conditions

Long-path echo is a salient factor that causes the degradation of the intelligibility of speech transmitted through a wide area outdoor environment or a very large indoor space using public-address systems. To robustly transmit speech information under such conditions, it is important to overcome this effect by controlling the characteristics of speech sounds. In this study, we consider the effects of inserting pauses between the words of a sentence. We performed word intelligibility tests using a series of four continuous words, called a quadruplet. Various pause lengths and long-path echo patterns were applied to the quadruplet. The results of the experiments demonstrate that word intelligibility under a long-path echo is significantly improved by the insertion of pauses between the words. Intelligibility can approach the same levels observed in the absence of echoes for a pause length of approximately 200 ms, which is almost the same as the length of 1-mora for the words used in the experiments. Moreover, this 200 ms pause is known to be sufficient to improve speech recognition in older adults. These results suggest that inter-word pauses of a length of approximately 1-mora can generally enhance the robustness of speech communication systems when used under a severe environment.

Experimental measurements of word intelligibility among pre-school children under differing conditions of reverberation and signal-noise ratio

Many children, from 0 to 5 years old, spend most of their active hours in child day-care centers. Rooms in these centers are often very noisy, due to both the voices of children or sounds of various activities and to the excessive reverberation attributable to the little use of sound-absorbing materials. Some researchers have studied the speech intelligibility of elementary school children, but we were unable to find those of pre-school children. Thus, in this study, an experiment was carried out a word intelligibility test on children from 3 to 5 years old in a classroom of a child day-care center. The procedure was like a game. Each of the presented words was mixed with pink noise at one of two S/N ratios, and convolved with the impulse responses of three reverberation times of 0.3, 0.9, and 1.8 s. Children were asked to judge whether the word was food or not, and to raise their hands holding yes or no signs. As a result, the correct answer ratio for 3-year-old children was lower than the ratio for other groups. Also, the correct answer ratio for 4- and 5-year-old children was slightly lower than control groups of elementary school pupils and college students.

Effects of dominance of the first-arriving sound on speech intelligibility in reverberant sound fields

Generally speaking, the energy of the direct sound exceeds that of reflections in rooms for listening to speech. In this case, early-reflections enhance loudness of the direct sound. It is well-known that the useful-to-detrimental sound ratio, in which useful is the sum of the direct sound and early-reflections and detrimental is late-reflections, can predict speech intelligibility. On the other hand, considering speech leakage from the adjacent room, the first-arriving sound is attenuated by the boundary wall, and it is sometimes too weak to trigger the temporal integration of loudness. This would change the time structure of the useful energy. In the present study, prediction of speech intelligibility in such particular situations was investigated. Word intelligibility tests were performed in reverberant sound fields varying reverberation time, direct-to-reverberation ratio, and rise time at the onset of reverberation sound. The results demonstrated that prediction accuracy of the useful-to-detrimental ratio was still in a practical range though the parameters systematically affect the prediction error.

The entropic brain: effortful speech reflcted in organisation of the network graph

Uncertainty of incoming information increases the amount of mental effort. Scientists might have been interested in quantifying the amount of this mental effort, as evidenced by the great amount of research devoted to computing the energy used to deal with the uncertainty of an event. One of these measures is Shannon’s entropy, which was initially used as a measure of the level of uncertainty with respect to the outcome of an event (Shannon, 1948). As defined by Shannon, entropy is expected to increase as the level of disorder (uncertainty or task complexity) increases. Therefore, to evaluate how the brain deals with entropy, one could either locate the focal area showing activity that is positively correlated with the degree of uncertainty or consider information transfer. We speculated that a functionally effective brain would modulate its network to fit to the varying entropy, since computation in the brain is probabilistic. Therefore, we put forward two hypotheses: 1) the functional brain architecture will reflect the varying entropy, and 2) the entropy will be modality-specific. In order to test these hypotheses, we estimated several network properties of the magnetoencephalography time-series signals obtained from healthy monolingual listeners. We particularly focused on whether entropy reflects the modality-specific processing load, especially when the degrees of the processing load of the two events were similar in terms of the rate of accuracy and the length of response time between tasks. In order to manipulate both entropy and modality, we varied the processing complexity (easy vs. difficult) in either a linguistic or non-linguistic (pitch change detection vs. word intelligibility test) task. Using graph-theoretical measures, the global organisation measures of the network, such as its small-worldness, correlation coefficient, global efficiency, and characteristic path-lengths, were compared within the network extracted as a set of 78 brain atlas nodes. The results showed a significant main effect of task complexity on the brain network properties, demonstrating that task load is indeed ubiquitous regardless of task modality. Equally importantly, we also found a pronounced task-specific difference in the network properties between linguistic and non-linguistic modalities. Regardless of modality, in the effortful tasks, the characteristic path-length and the correlation-coefficient were significantly larger, whereas the linguistic tasks resulted in significantly higher small-worldness, with the hubs located at the usual language nodes. Our findings collectively suggest that task load is ubiquitous but is also modality-specific in the brain network properties, as evidenced by the specific network graph measures.

主観的評価に基づくオープンプランオフィスにおける執務作業時のサウンドマスキングの適用に関する検討

The purpose of this study was to examine masking systems between desks in open plan offices installed with an appropriate sound masking system. Word intelligibility tests for unwanted speech interference from nearby workstations, and subjective judgements on the degree of discomfort of masking sound were conducted in a simulated office workplace. The results were as follows: (1) The level of masker sound could be decreased to 5-10 dB when focused on desk work relative to the masking level in non-desk work; (2) We showed that the optimum masking level in a desk work environment is a target/masker ratio of -2 dB.

Swedish Test of Intelligibility for Children (STI-CH) – Validity and reliability of a computer-mediated single word intelligibility test for children

The aim of this study was twofold: first, to describe a new Swedish intelligibility test (Swedish Test of Intelligibility for Children, STI-CH) and second to evaluate its validity and reliability. STI-CH is based on the repetition of single words. Ten children with a speech-sound disorder (4:6–8:3 years of age, mean = 6.0 years) and 10 children with typical speech and language development (4:8–7:4 years of age, mean = 5.9 years) were included. Twenty speech-language pathology students served as listeners. Intra-judge reliability was high (r > 0.92), as was the intra-class correlation of inter-judge reliability (0.97). In terms of validity, there was a significant difference in STI-CH scores between the two groups, and the scores correlated statistically significantly with the Percentage of Consonants Correct (r = 0.94) and with intelligibility in spontaneous speech (r = 0.85). To sum up, the results indicate that STI-CH could be an option for the assessment of intelligibility in Swedish-speaking children, and that the principles used in the development of the test could be of use in the design of intelligibility tests in languages other than Swedish.

Single-word intelligibility testing in aphasia: Alternate forms reliability, phonetic complexity and word frequency

Background: We recently constructed a single-word speech intelligibility test intended to quantify magnitude of speech sound impairment in individuals with aphasia. To minimise listener learning and strategising, the test included a large set of possible (alternate) test forms and was constructed with high and diverse phonologic similarity among candidate words. Although the corpus was limited in phonetic complexity to single syllables, and although criteria for minimal word frequency were applied, it is possible that more fine-grained differences in phonetic complexity and/or word frequency may introduce varying levels of difficulty across alternate test forms and that reliability, therefore, may be compromised.Aims: The dual purposes of this study were to evaluate alternate forms reliability for the new intelligibility test and to determine whether word frequency and/or phonetic complexity affected word identification scores.Methods & Procedures: All words in the 600-word test corpus were coded for overall phonetic complexity and for frequency of occurrence in spoken English. Fifty-one versions of the target 50-word test were generated from this corpus by following designated pseudo-random selection procedures. Speech samples were collected from 13 speakers with aphasia, who each repeated three or four of these 50-word sets. Ten normal-hearing listeners were asked to indicate the words they thought the speakers were trying to say. Per cent accuracy was computed for each speaker, test form and target word.Outcomes & Results: The intra-class correlation within speakers was 0.97, indicating that the scores across alternate forms were highly reproducible. Analyses at the word level showed that both phonetic complexity and word frequency affected identification accuracy. The effects were seen in almost all participants whose intelligibility scores were in the impaired range.Conclusions: High frequency of occurrence and low phonetic complexity increase repetition accuracy for individuals with moderate to severe sound production difficulties and aphasia. However, the pseudo-random word selection in the examined word intelligibility test was sufficient to ensure strong alternate forms reliability. Further constraints on the word selection process for this test are not warranted.

Estimation of speech privacy performance from acoustic parmeters in two adjacent rooms

Sato et al. suggested the equal-intelligibility contours that enable us to predict sound insulation performance and background noise level required to achieve a certain level of speech privacy/security on the basis of the word intelligibility test [Sato et al., Appl. Acoust. 73, 43-49 (2012)]. However, temporal aspects of sound fields, which would affect intelligibility scores, were not considered in the intelligibility test. In the present study, intelligibility tests were performed to investigated the effects of temporal aspects of sound fields on the scores. The possibility of using reverberation time as one of the predictors of speech privacy performance will be discussed on the basis of intelligibility tests.

Relationship between sound insulation performance of walls and word intelligibility scores

Conversations must be shielded from people in an adjacent room if they include confidential information. Word intelligibility tests were performed in a total of 185 sound fields to examine the relationship between sound insulation performance and the degree of conversation leakage. The parameters of the test sound fields were background noise level in the adjacent room and the level difference between the two rooms. The background noise level was varied from 30 to 50 dB (A-weighted). The level difference was parametrically varied in terms of eight frequency characteristics and 10 absolute values. The results showed that word intelligibility scores were strongly correlated with the A-weighted speech-to-noise ratio and SNR uni32 . Equal-intelligibility contours, which can easily show the weighted level difference and A-weighted background noise level required to achieve a certain level of word intelligibility scores, were obtained from a multiple logistic regression analysis.

Effect of speed difference between time-expanded speech and moving image of talker’s face on word intelligibility

This study investigated effects of asynchrony between speech signal and moving image of talker’s face induced by time-expansion of the speech signal on speech intelligibility. Word intelligibility test was performed to younger listeners. Japanese 4-mora words were uttered by a female speaker. Each word was processed with STRAIGHT software to expand the speech signal by from 0 to 400 ms. These signals were combined with moving image of talker’s face which was kept at original speed. This test was performed under three conditions: visual-only, auditory-only, and auditory-visual (AV) condition. Results showed that intelligibility scores under AV condition were statistically higher than those under auditory-only condition even when the speech signal was expanded by 400 ms. These results suggest that moving image of talker’s face is effective to enhance speech intelligibility if the lag between the speech signal and moving image of talker’s face does not exceed 400 ms.

Critical‐band compression method of speech enhancement for elderly people: Investigation of syllable and word intelligibility

Auditory filters for the hearing impaired tend to be wider than those of normal hearing people. Thus, the frequency selectivity decreases because of increased masking effects [Glasberg and Moore, J. Acoust. Soc. Am., 79(4), 1020‐1033, 1986]. We have developed a method, called compression, in which the critical band is compressed along the frequency axis [Yasu et al., Handbook of the International Hearing Aid Research Conference (IHCON), 55, Lake Tahoe, 2004]. We investigated whether our method improves syllable and word intelligibility. Thirty one elderly people participated in experiments. First, we measured the auditory filter bandwidth using a notched noise method [Patterson, J. Acoust. Soc. Am., 59(3), 640‐654, 1976]. Second, we conducted syllable and word intelligibility tests. The compression rates of critical‐band compression were set to 0% for the original, and 25%, 50%, and 75%. The results were that the percentages of correct responses were almost the same at 0%, 25% and 50% compression rates for syllable and word intelligibility. A significant correlation was not obtained between the compression rate of processing and the auditory filter bandwidth. [Work supported by JSPS.KAKENHI (16203041) and Sophia University Open Research Center from MEXT.]

Development of word lists for word intelligibility tests in Japanese

Speech intelligibility tests are widely used to evaluate personal speech hearing ability accurately. Although various lists have been proposed for the tests, almost all lists in Japan do not properly consider word difficulty as an aspect of speech recognition. Consequently, even if these tests are performed under identical conditions, correct results cannot be obtained. We have been developing word lists for word intelligibility tests to cope with these problems. We control word familiarity, which strongly affects word recognition scores, to equalize difficulty in recognition between word lists. Words are divided into four word-familiarity ranks. Twenty lists of 50 words are constructed in each rank by considering phonetic balance. Moreover, to compensate slight differences in intelligibility scores among word lists within the same word-familiarity rank, the relationship between the difference of word familiarity and signal-to-noise ratio (SNR) is estimated. Then this relationship is used to compensate the difference of the intelligibility score. The other compensation method is based on speech recognition threshold (SRT). The intelligibility score of each word can be equalized by changing the SNR according to the difference of the SRT of the word. The effectiveness of these compensation methods is discussed.

Effects of Instantaneous Multiband Dynamic Compression on Speech Intelligibility

The recruitment phenomenon, that is, the reduced dynamic range between threshold and uncomfortable level, is attributed to the loss of instantaneous dynamic compression on the basilar membrane. Despite this, hearing aids commonly use slow-acting dynamic compression for its compensation, because this was found to be the most successful strategy in terms of speech quality and intelligibility rehabilitation. Former attempts to use fast-acting compression gave ambiguous results, raising the question as to whether auditory-based recruitment compensation by instantaneous compression is in principle applicable in hearing aids. This study thus investigates instantaneous multiband dynamic compression based on an auditory filterbank. Instantaneous envelope compression is performed in each frequency band of a gammatone filterbank, which provides a combination of time and frequency resolution comparable to the normal healthy cochlea. The gain characteristics used for dynamic compression are deduced from categorical loudness scaling. In speech intelligibility tests, the instantaneous dynamic compression scheme was compared against a linear amplification scheme, which used the same filterbank for frequency analysis, but employed constant gain factors that restored the sound level for medium perceived loudness in each frequency band. In subjective comparisons, five of nine subjects preferred the linear amplification scheme and would not accept the instantaneous dynamic compression in hearing aids. Four of nine subjects did not perceive any quality differences. A sentence intelligibility test in noise (Oldenburg sentence test) showed little to no negative effects of the instantaneous dynamic compression, compared to linear amplification. A word intelligibility test in quiet (one-syllable rhyme test) showed that the subjects benefit from the larger amplification at low levels provided by instantaneous dynamic compression. Further analysis showed that the increase in intelligibility resulting from a gain provided by instantaneous compression is as high as from a gain provided by linear amplification. No negative effects of the distortions introduced by the instantaneous compression scheme in terms of speech recognition are observed.

Using listening difficulty ratings of conditions for speech communication in rooms

The use of listening difficulty ratings of speech communication in rooms is explored because, in common situations, word recognition scores do not discriminate well among conditions that are near to acceptable. In particular, the benefits of early reflections of speech sounds on listening difficulty were investigated and compared to the known benefits to word intelligibility scores. Listening tests were used to assess word intelligibility and perceived listening difficulty of speech in simulated sound fields. The experiments were conducted in three types of sound fields with constant levels of ambient noise: only direct sound, direct sound with early reflections, and direct sound with early reflections and reverberation. The results demonstrate that (1) listening difficulty can better discriminate among these conditions than can word recognition scores; (2) added early reflections increase the effective signal-to-noise ratio equivalent to the added energy in the conditions without reverberation; (3) the benefit of early reflections on difficulty scores is greater than expected from the simple increase in early arriving speech energy with reverberation; (4) word intelligibility tests are most appropriate for conditions with signal-to-noise (S/N) ratios less than 0 dBA, and where S/N is between 0 and 15-dBA S/N, listening difficulty is a more appropriate evaluation tool.

Complementary relationship between familiarity and SNR in word intelligibility test

Complementary relationship between familiarity and SNR in word intelligibility test