Download the_brain_model_of_intelligibility_in_business_telephony_1.pdf

Contents : The "BRAIN" Model of Intelligibility in Business Telephony Jeff Rodman Fellow/CTO September 2006 "I am in susquehanna jail. Susquehanna. S u s q 'Q'! 'Q ' you know the thing you play billiards with. Billiards. B i l l i No 'L' for 'Larynx!' L a r y n No! Not 'M'! 'N'!!! 'N' as in 'Neighbor'! N e i ..." ---Eric Blore in jail and on the phone to Fred Astaire in "Shall We Dance " 1937 Introduction There are times when there is no substitute for being understood on the telephone. Well before 1937 the limitations of the telephone in accurately conveying speech were known. Irregular and limited bandwidth noise variations in end-to-end loudness sidetone and distortion had all been identified as contributors to degradation of the spoken word. In 1910 Campbell performed experiments in which he found 59 percent accuracy when words were called over the telephone as compared to 96 percent through open air. The abilities of the telephone as an efficient and accurate channel for human speech have always been regarded with a bit of a wink and a chuckle tolerated due to a common understanding that it is the best available. Despite modern digital trunking and switching technology this remains an everyday problem. Why Analog loop lengths building wiring variable line equalization characteristics poor handset and speakerphone designs mixed networks noise in conference rooms paper shuffling pen tapping fan noise and a host of other issues are still with us even here in the digital age. This paper discusses these issues and presents the "BRAIN" model of critical elements in business telephony. We show how their mutual dependencies can be used to improve telephone and audio system performance and how contemporary systems can produce direct benefits to clear communications. What Distinguishes Business Telephony In business the critical role of the telephone is magnified for a number of reasons. Consider the following characteristics of business telephony. Time is often in more demand than in personal telephony which makes misunderstandings and "can you repeat that " more frustrating. Meetings have 2 fixed lengths so lost time is irretrievable. The nature of discussions makes accuracy critical so the many ways that telephones distort speech and degrade accuracy carry a real cost to business. Users are often talking to people that they do not know and have never met before. A phone call is the first impression for both parties in a high-value relationship. Phone calls often occur between people who have different native languages or dialects so accented speech is added to the other burdens of telephony. Conferences occur among groups which increases both the cost and the potential value of the meeting due to the number of participants and the difficulty of scheduling them. Yet to accommodate such a group the quality of the sound must be degraded because some kind of speakerphone is required. This can introduce room reverberation fan noise clipping and feedback interruptions and multiple participants sitting near and far away. Having multiple talkers participating in a conference makes fast and accurate identification of who is talking more important but also more difficult. Meetings can be very long yet require sustained attention. This puts an increased strain on sound quality because small differences in speakerphone performance add up to big differences in fatigue and attention. Intelligibility and the BRAIN Model For all these reasons intelligibility (how easily speech is understood) is especially critical in business telephony. Yet perversely the challenges to intelligibility are magnified in the group settings that are so common in business. Let us look at the components of speech intelligibility. We hear and understand speech in three main stages which are called here the physical cognitive and analytical. In the physical stage speech is carried from the talker's mouth to the listener's ears and the fidelity with which this is done is paramount. In the cognitive stage the listener resolves ambiguities in what she's heard by applying simple reasoning such as grammatical and accent rules to the local context of the word. And finally in the analytical stage those words that are not resolved through the first two stages are subjected to more intense scrutiny examining the troubling words in broader contexts to see if their identities can be inferred. These second two layers of comprehension are increasingly distracting. Additionally 3 there are instances such as speech from foreign talkers where the assumptions of these contextual analyses do not hold: a person who does not share the same native language or dialect will likely use entirely different words or sounds than those that are expected. Figure 1. The BRAIN Model of Physical Speech Communication For these reasons the p

INFO HASH : a9ca3a6277e9e27ea097af97e24918f08f9f7f01

Download now

File Size: 98 kb