Predicting speech information from the audiogram ...

Predicting speech information from the audiogram and vice versa

Introduction: The prediction of speech perception scores from the audiogram has been studied often in the past with methods such as the articulation index [1,2], the speech intelligibility index [3], and by showing speech sounds on the audiogram as in Figure 1. It is rare that the audiogram has been predicted from a speech perception test. One of the Figure 1. Audiogram showing the reasons for this is that it is approximate intensities and frequencies difficult to extract frequencyof selected speech sounds. specific information from speech perception data. The Infogram™ described here is an attempt to do just that, and to display the speech information in a way that is intuitive for lay people to understand. The Infogram™ is loosely based on the research of Miller and Nicely [4] who found relationships between the information transmitted for a set of speech features and the cut-off frequencies of low- and high-pass filters. The Infogram™: The Infogram™ is derived from the Speech Perception Test (SPT) found at http://apps.blameysaunders.com.au/wordtest/

Audiogram PCA coefficients 0.6 0.4 0.2

PC2a

0 250

-0.2

500

750

PC3a

1000 1500 2000 3000 4000 6000

PC4a

-0.4 -0.6

Normal (5 dB HL) Mild (20 dB HL) Moderate(40 dB HL) Severe (70 dB HL) Profound (90 dB HL) Corner Ski-slope

Figure 4b. Infograms™ for flat hearing losses tended to slope downwards from left to right as expected from the higher intensity of speech at low compared to high frequencies. The ski-slope audiogram corresponded to a more steeply sloping Infogram™ and the corner audiogram predicted very poor speech perception.

Hypothetical Infograms™

Audiometric frequency (Hz)

Figure 3. The first four principal components accounted for 95% of the variability in a sample of 2534 best audiograms

The first component, PC1a, was highly correlated with the Pure Tone Average hearing loss (r = 0.938, p < 0.001). PC2a and PC3a gave measures of slope and convexity of the best audiogram respectively. Infogram™ data collection and analysis: 6079 de-identified SPT results were collected clinically or online and Infograms™ were calculated. Outliers were excluded, leaving 6068 Infograms™. Principal components analysis was used to extract shape information from the Infograms™. The first 5 principal components accounted for 95% of the variability in this sample of Infograms™. As for the audiogram, the first component, PC1i, was highly correlated with PTA (r = -0.643, p < 0.001). Multiple regression analyses There were 408 clients for whom both audiogram and Infogram™ were available within the two data sets. The SPT was performed binaurally in a free field at a comfortable level, so it was assumed that the results would reflect the hearing in the better ear, or the best audiogram. The better of the left and right ear thresholds was chosen at each frequency to determine the best audiogram. Multiple linear regression was used to generate matrices relating the best audiogram principal components to the Infogram™ principal components and vice versa. These matrices were combined with the principal component calculation matrices to produce estimates of the audiogram from the Infogram™ and vice versa.

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

flat80% flat60% flat40% flat20% flat10% sloping

Figure 5a. Hypothetical Infograms™ to illustrate the correspondence between audiograms and Infograms™

Speech audiograms estimated from Infograms™ 250

500

Frequency (Hz) 750 1000 1,500 2000 3000 4000 6000

0 10 20 30 40 50 60 70 80 90

flat80% flat60% flat40% flat20% flat10% sloping

Figure 5b. Speech audiograms predicted from the flat Infograms™ tended to slope downwards from left to right while the sloping Infogram™ predicted a skislope audiogram

Results: The predicted hearing thresholds in the better ear, based on the information transmission results from a single SPT in the binaural condition at 65 dB SPL were highly correlated with the actual hearing thresholds measured with an audiometer (p < 0.001). Frequency Correlation coefficient

250

500

750

1000

1500

2000

3000

4000

6000

0.48

0.512

0.579

0.608

0.68

0.702

0.648

0.628

0.565

Similarly, the information transmission values for the speech features predicted from the audiogram were highly correlated with the actual values derived from the SPT (p < 0.001). Speech feature Correlation coefficient

Nasality

vHeight

Manner

Voicing

Contour

vPlace

vLength

0.601

0.615

0.678

0.630

0.634

0.667

0.653

Affric’n Sibilance

Place

0.698

0.715

0.658

Hearing Thresholds (dB HL)

-15 100

Frequency (Hz) 1000

5 25 45 65 85

10000 Normal (5 dB HL) Mild (20 dB HL) Moderate(40 dB HL) Severe (70 dB HL) Profound (90 dB HL) Corner Ski-slope

105 125

Figure 4a. Hypothetical audiograms to illustrate the correspondence between audiograms and Infograms™.

References: [1] Kryter, Karl D. (2005) "Methods for the calculation and use of the articulation index." The Journal of the Acoustical Society of America 34.11: 1689-1697. [2] French, N. R., and J. C. Steinberg. (2005) "Factors governing the intelligibility of speech sounds." The journal of the Acoustical society of America 19.1: 90-119. [3] American National Standards Institute. American National Standard: (1997) “Methods for Calculation of the Speech Intelligibility Index.” Acoustical Society of America. [4] Miller GA, Nicely PA. (1955) “An analysis of perceptual confusions among some English consonants.” J Acoust Soc Am, 27, 338-352. [5] Blamey, Peter J. (2012) “Alternatives to the audiogram for hearing aid fitting.” IHCON, August 2012, Lake Tahoe.

Figure 6. These data are for the client whose Infogram™ is shown in Figure 2. The “speech audiogram,” shown in green, was very close to the right (better ear) audiogram but slightly overestimated the low and mid-frequency losses.

Conclusions: Infograms™ provided hearing information that was equivalent to the conventional audiogram in the better ear. “Speech audiograms” derived from binaural Infograms™ were highly correlated with conventional audiograms in the better ear and Infograms™ predicted from conventional audiograms were highly correlated with actual Infograms™. An earlier attempt at this study failed because only 10 items were used in the speech test and this resulted in poor repeatability [5]. The 50-word SPT overcame this problem. Recommendations and Caveats:

Hypothetical audiograms

The goal of this study was to investigate methods of converting between audiograms and Infograms™ and demonstrate their equivalence or otherwise. Our reasons for attempting this study were to increase the understanding of the connection between speech perception performance and hearing loss, and to remove one of the barriers to hearing aid use, namely the need for a hearing test conducted with an audiometer in a sound booth.

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

PC1a

To illustrate the correspondence, flat audiograms for normal hearing, mild, moderate, severe and profound hearing losses, and corner and ski-slope audiograms are shown in Figure 4a below, and the corresponding Infograms™ are shown in Figure 4b to the right. Hypothetical Infograms™ and the corresponding “speech audiograms” are shown in Figures 5a and 5b.

Figure 2. A typical Infogram™ showing the percentage of information transmitted for ten speech features ordered from those conveyed by low frequencies on the left to those conveyed by high frequencies on the right. The corresponding audiogram is shown in Figure 6

Infograms™ estimated from hypothetical audiograms Information transmission

Methods: Audiometric data collection and analysis: 2534 de-identified audiograms from online and clinic customers of Blamey & Saunders Hearing Pty Ltd were collected and principal components analysis (PCA) was used to extract shape information.

Information transmitted

Abstract: Aims: Information transmission analysis of vowel and consonant confusions from a monosyllabic word test can be used as an alternative to the audiogram for measurement of hearing loss, for hearing aid fitting, and for counselling people who have hearing difficulties about listening and communication strategies and tactics. The goal of this study was to investigate methods of converting between audiograms and Infograms™ and demonstrate their equivalence or otherwise. Methods: A CVC word test, (Speech Perception Test, SPT) and conventional audiometry were used with 408 people attending an audiology clinic. Information transmission analysis of vowel and consonant features was applied to the individual word test results (the Infogram™). Principal Components Analysis and Multiple regression were used to develop formulae to convert from the Infogram™ to the audiogram and vice versa. Results: There was a high correlation between the conventional audiograms and the “speech audiograms” derived from the Infogram™ at every frequency (p < 0.001). Similarly, there was a high correlation between the measured information transmitted and the information transmitted estimated from the audiogram for every speech feature (p < 0.001). Although the “speech audiograms” followed the same general shape as the conventional audiograms, they tended to be smoother and have shallower slope because of the broadband nature of speech sounds. Conclusions: The relationships and equivalence between word test results and audiograms will enable measurement of hearing and hearing aid fitting without specialised equipment or expertise. This is of particular relevance in regions with limited availability of audiologists, audiometers, sound booths, etc. It also improves self-fitting of hearing aids.

Hearing loss (dB HL)

Peter J Blamey Jeremy K Blamey Daniel Taft 2 Elaine Saunders 2

Institute, www.bionicsinstitute.org 2 Blamey and Saunders Hearing Pty Ltd, www.blameysaunders.com.au 3 Department of Audiology and Speech Pathology, The University of Melbourne, www.medoto.unimelb.edu.au 4 Department of Medical Bionics, The University of Melbourne, 384-388 Albert Street, East Melbourne, Victoria 3002, Australia

2

Coefficient

1,2,3,4

1 Bionics

For the best results, the SPT should be performed under controlled conditions: • If the SPT is presented at a level greater than 65 dB A, the “speech audiogram” may underestimate the hearing loss. • If the SPT is presented at a level less than 65 dB A or in poor signal-to-noise ratio, the “speech audiogram” may overestimate the hearing loss. • When the SPT is performed in free field with binaural listening, the “speech audiogram” represents the estimated hearing thresholds in the better ear. Monaural thresholds may be estimated with monaural presentation under headphones. • The current SPT is designed for native speakers of Australian English with a full adult vocabulary. Reduced performance has been observed for non-native English speakers and for people with non-Australian accents. American, New Zealand, and Oxford English versions are under development. Hearing aid fittings have been performed directly from the Infogram™ and results will be presented in Oral Session FP25.

Acknowledgments: Infogram™ is the trademark of Blamey & Saunders Hearing Pty Ltd. The SPT is freely available to the public at http://apps.blameysaunders.com.au/wordtest/. The study was funded by Blamey & Saunders Hearing Pty Ltd. The Bionics Institute acknowledges the support it receives from the Victorian Government through its Operational Infrastructure Support Program.