Articulation index
The articulation index ( AI ) describes the relative contributions of the frequency groups to speech intelligibility .
It turns out that the relative contributions to speech intelligibility in the range up to approx. 800 Hz are negligibly small. Only the ranges from 1,600 to 3,500 Hz are decisive for language discrimination .
The articulation index was developed around 1950 as part of studies of speech intelligibility on the telephone in the USA. It was firmly defined at ANSI . The AI is a relatively uncomplicated value to determine. For this you only need a hearing threshold and add up the points that are audible, divide the sum by x and get a value.
A value between 0 and 1 is always calculated. There are tables for this purpose in which the AI is shown depending on different language materials. There you can see that with an AI of 0.3, more than 90% sentence comprehension can be achieved. An improvement in AI does not always go hand in hand with an improvement in intelligibility.
The AI primarily describes the understanding of speech in people with normal hearing. However, in connection with the hearing impaired and the improvement of speech intelligibility through hearing aids, this value has long since reached the limits of its informative value .
The SII (Speech Intelligibility Index)
The SII is a value that describes speech intelligibility. The SII is a continuation of the AI.
Background: In the ANSI , the AI was replaced by the SII. In contrast to AI, the factors to be considered are many times more complex. To calculate the SII, the importance of speech intelligibility was determined in up to 21 frequency bands and given an individual value depending on the frequency. In addition, the properties of the inner ear (e.g. the upward masking [overlaying of high frequencies by too many low frequencies] and the decreasing frequency resolution of the cochlea at higher levels) are taken into account.
The SII can be given in% or also from 0 to 1. It is interesting that as the speech level increases, the AI continues to grow, but the SII decreases at some point. This is an interesting value, especially in hearing aid fitting, for people with moderate to profound hearing loss.
Summary of meaning:
The SII is a value that, in contrast to the older AI, has a predictive character.
An SII of 0.5 or 50% does not mean that 50% speech intelligibility has been achieved. An SII of 0 means that no speech information can be used, an SII of 1 or 100% means that all speech information can be used. If the SII increases, the speech intelligibility also increases.
Acceptable noise level
The Acceptable Noise Level (ANL) test is a measure of the background noise level that a person is willing to tolerate. In recent years, the ANL has gained interest among researchers and hearing care professionals as it has the ability to predict with 85% accuracy who will be successful with a hearing aid.
Performing the ANL test is relatively quick and easy. First, fluent speech is presented to the listener through headphones. With an adaptive approach, the listener is first instructed to set the speech level to a value that is “too loud”, then “too quiet” and then “most pleasant”. Next, a background noise is added, usually a babble of voices with many speakers, and the listener is instructed to adjust the noise level, first to a value that is “too loud to understand speech”, then to a value that is “quiet is enough to understand language ”and finally to the highest value that the subject would“ come to terms with while following the language ”. The difference between the listener's MCL and the maximum tolerated background noise BNL is the ANL. The test takes 2-3 minutes to complete.
A lower ANL value reflects a higher tolerance for background noise.
There are three different ANL categories: low, medium, high.
People who have “low” ANLs (less than 7 dB ) are usually successful hearing aid users.
People with “high” ANLs (greater than 13 dB) are generally not successful hearing aid users.
“Medium” ANLs (7–13 dB) 50:50 chance (according to Nabelek et al.).
Nabelek et al. Showed that most of the hard of hearing people had ANLs between 0 and 25 dB. The most common values were 10–11 dB.
ANLs do not appear to be related to a person's age, gender, or sensitivity to hearing, or a preference for the presence of background noise. It is currently unclear whether the ANLs are related to how well a person understands speech - some researchers suggest that ANLs and speech intelligibility do not always correlate, while other researchers suggest that people with better speech intelligibility also have lower ANLs. Similarly, studies examining ANLs with and without a hearing aid have produced conflicting results, with Nabelek et al. Showed that ANLs are independent of the test conditions and Ahlstrom found that ANLs with a hearing aid are lower than ANLs without a hearing aid.
In addition to these results, both directional microphones and noise reduction technologies have shown that they improve the ANLs of the hearing aid wearer by approximately 2.5-4 dB compared to hearing aids that did not use this feature. These results are very interesting because they indicate that hearing aid settings and the signal processing of hearing aids enable people to tolerate higher levels of background noise and thus improve the success rate of hearing aid users. Also, if we could understand the traits people use to determine their tolerance for background noise, that information could lead to an understanding of who is most likely to benefit from these technologies.
Since the ANL instructions require the audience to be able to follow the main speaker, it is possible for some people to adjust the level of background noise according to the criterion of intelligibility. If so, we suspect that these people are more likely to benefit from directional mics or other SNR-enhancing technology than listeners whose ANLs are based on other criteria.