Intelligibility Of Natural Speech Research Articles

Visemic processing in audiovisual discrimination of natural speech: A simultaneous fMRI–EEG study

In a noisy environment, visual perception of articulatory movements improves natural speech intelligibility. Parallel to phonemic processing based on auditory signal, visemic processing constitutes a counterpart based on “visemes”, the distinctive visual units of speech. Aiming at investigating the neural substrates of visemic processing in a disturbed environment, we carried out a simultaneous fMRI–EEG experiment based on discriminating syllabic minimal pairs involving three phonological contrasts, each bearing on a single phonetic feature characterised by different degrees of visual distinctiveness. The contrasts involved either labialisation of the vowels, or place of articulation or voicing of the consonants. Audiovisual consonant–vowel syllable pairs were presented either with a static facial configuration or with a dynamic display of articulatory movements related to speech production.In the sound-disturbed MRI environment, the significant improvement of syllabic discrimination achieved in the dynamic audiovisual modality, compared to the static audiovisual modality was associated with activation of the occipito-temporal cortex (MT+V5) bilaterally, and of the left premotor cortex. While the former was activated in response to facial movements independently of their relation to speech, the latter was specifically activated by phonological discrimination. During fMRI, significant evoked potential responses to syllabic discrimination were recorded around 150 and 250ms following the onset of the second stimulus of the pairs, whose amplitude was greater in the dynamic compared to the static audiovisual modality. Our results provide arguments for the involvement of the speech motor cortex in phonological discrimination, and suggest a multimodal representation of speech units.

Natural Speech Intelligibility in Theatres in Relation to Their Acoustics

There is a certain tendency in the design of theatres to make the halls quite large. From a perspective of natural speech intelligibility and strength of speech this is disadvantageous, because an actor's voice has a certain, limited loudness and consequently the signal-to-noise ratio at the listener may become too low. Based on the influence of signal/noise ratio on speech intelligibility, it is deduced that the strength G ≥ 6 dB and room volumes have to be limited to 4000–4500 m3 in order to maintain sufficient loudness for natural speech. Sound level measurements during performances with natural speech in a theatre have been performed, to determine background noise levels in the hall due to the audience and to investigate the signal-to-noise ratio of the actors voice at the audience. The background levels are mainly determined by installation noise and not by the influence of the audience.

Audiovisual asynchrony detection in human speech.

Combining information from the visual and auditory senses can greatly enhance intelligibility of natural speech. Integration of audiovisual speech signals is robust even when temporal offsets are present between the component signals. In the present study, we characterized the temporal integration window for speech and nonspeech stimuli with similar spectrotemporal structure to investigate to what extent humans have adapted to the specific characteristics of natural audiovisual speech. We manipulated spectrotemporal structure of the auditory signal, stimulus length, and task context. Results indicate that the temporal integration window is narrower and more asymmetric for speech than for nonspeech signals. When perceiving audiovisual speech, subjects tolerate visual leading asynchronies, but are nevertheless very sensitive to auditory leading asynchronies that are less likely to occur in natural speech. Thus, speech perception may be fine-tuned to the natural statistics of audiovisual speech, where facial movements always occur before acoustic speech articulation.

Intelligibility enhancement of synthetic speech heard via telephone in a noisy environment

The intelligibility of synthetic speech, like the intelligibility of natural speech, decreases if it is transmitted via telephone and/or heard in a noisy environment. A natural speaker adapts his speaking style to these hard speaking conditions in order to restore intelligibility. With new applications for speech synthesis, e.g., navigation systems for cars, the need for adapting synthetic speech to hard speaking conditions arises. Here, the results of intelligibility tests for German databased synthetic speech in comparison to natural speech are presented, and first methods of intelligibility enhancement for synthetic speech are given. It is shown that under the conditions mentioned above unadapted synthetic speech has a lower segmental intelligibility for all phonemes or phoneme clusters than natural speech. For that reason the methods applied affected all speech sounds equally. They are based on prevoius studies of natural speech produced under stress conditions like a noise afflicted environment [Summers et al., J. Acoust. Soc. Am. 84, 917–928 (1988)]. It is shown that the intelligibility of synthetic speech is improved significantly by these enhancement methods. The results give evidence to an individual adaptation for phonemes or phoneme clusters in future.

Parameters governing speech intelligibility

Intelligibility of natural speech is talker-dependent and the sources of the differences are unclear. Speech intelligibility can be enhanced by mode of talking, one of which has been termed “clear speech”. This involves talking in such a manner as to improve hearing-impaired listeners' understanding of speech and is characterized by a more precise articulation rather than just a slow rate of talking. Some of the parameters which have been identified in clear speech were subsequently used with some success in speech enhancement systems for improved speech intelligibility for hearing-impaired listeners. Understanding of parameters governing speech intelligibility is very important for construction of speech perception tests and for production of synthetic speech. Proliferation of synthetic messages in public areas demands that synthetic productions be highly intelligible.

A comparison of intelligibility among natural speech and seven speech synthesizers with listeners from three age groups

This investigation examined the intelligibility of natural speech and the synthetic speech produced by the Smoothtalker 3.0 male, Smoothtalker 2.0 male, Smoothtalker 2.0 female, RealVoice female, Attic R65B, Votrax SC02, and Lightwriter voices. All of these are commonly used in portable communication devices currently available on the commercial market. Both single word and sentence intelligibility across listeners from three age groups (7–8 year olds, 11–12 year olds, and adults) were assessed using tasks presented with no contextual cues. The results revealed Age Group X Voice interactions for both the word and sentence tasks, indicating different patterns of intelligibility across the age groups. The individual patterns, their implications, and future research directions arising from this study are described in detail.

Quality assessment of synthetic speech using word intelligibility scores

The quality of Japanese speech synthesized by rules cannot be sufficiently evaluated using only the 100‐monosyllable list that has been used for speech articulation tests: Word intelligibility must also be assessed. Word intelligibility tests on both natural and synthetic speech were conducted to clarify the differences between synthetic and natural speech in learning effects, the effects of external noise, and word familiarity. Both trained and untrained listeners were used. Each word list of 155 words, selected by our previously developed method, reflected the most important word attributes of Japanese: word length, word familiarity, and initial phoneme occurrence. The experimental results show that the intelligibility scores for synthetic speech with untrained subjects were fairly low and independent of noise level. These were greatly improved by training, depending on the noise level. However, natural speech intelligibility depends little on training. The results also show that the familiarity of the test words and the first phoneme in the word are important factors in recognizing Japanese words in both synthetic and natural speech.

A Comparison of the Aids Sentence List and Spontaneous Speech Intelligibility Scores for Dysarthric Speech

The aim of this study was to determine whether the Assessment of Intelligibility of Dysarthric Speech (AIDS) (Yorkston & Beukelman, 1981) sentence scores are realistic indications of the spontaneous speech performance of dysarthric speakers. Twenty mild-to-moderately dysarthric subjects were tape-recorded during the AIDS sentence task, as well as for a spontaneous speech sample. Subsequently, each sample was transcribed by eight untrained judges. Four scores were obtained from each transcription. T-tests were used to determine if there were significant differences between the scores for the two tasks. The results showed that the percentage intelligibility of the AIDS sentences was significantly higher than that of spontaneous speech; indicating that the AIDS scores may not be representative of the speakers' natural speech intelligibility. The AIDS sentence and spontaneous speech scores — speaking rate, rate of intelligible speech, and communications efficiency ratio, were not significantly different. Reli...

Some comparisons of intelligibility of synthetic and natural speech at different speech‐to‐noise ratios

Despite the extensive literature on the intelligibility of natural speech in noise, relatively little is currently known about the deleterious effects of noise on the perception (i.e., recognition, identification, and understanding) of synthetic speech. In this paper, we wish to report the results of a study that examined the effects of noise on the intelligibility of synthetically produced monosyllabic words using the traditional Modified Rhyme Test to measure intelligibility. Over a range of S/N ratios, we found that synthetic speech suffers a greater, decrement in performance than naturally produced speech. Moreover, the decrement in performance was even greater when the MRT response format was changed from a closed six‐alternative forced choice procedure to an open free response format. The increase in uncertainty affected recognition of the synthetic items more than the natural ones. These findings have implications for the use of voice response systems that employ synthetic speech in noisy listening environments where intelligibility of the linguistic message may be at a premium. [Supported by a research grant from NINCDS to Indiana University.]