Back to the Aural
Rehabilitation Class Page
I. Long Term Characteristics
II. Short Term Characteristics
III. Ways to Represent Speech
IV. Effects of Hearing Loss
V. Evaluation of Speech Recog
I. Long Term Characteristics
1. Average Level of Speech
Long-term Average Speech
________Greatest energy at ________ Hz; 6 dB/Oct slope above that
Characteristics.......
________ of speech is below 1000 Hz 95% of the ________ of speech is above 500 Hz Usefulness of ________ spectrum95% of the
Determine the amount of ________ neededDetermine audibility of speech signal
II. Short Term Characteristics
| ________ Long Duration ________ Strong Intensity Voiced ________ |
Aperiodic Short Duration High Frequency ________ Voiced/Voiceless ________ |
Perceptual Cue: Formants-peaks of energy at certain frequencies which relate to ________ positionMost Intense
________
________ Second peak determined by ________ placement of tongueFirst peak determined by tongue
Only need first two formants for perception
Relationship of F1 to F2
Consonants
Least Intense
________
Perceptual Cue: Duration and Frequency
relate to manner, place, and voicing of articulation
Speech Banana
Frequency and Intensity of Speech Sounds
Useful for ________Dynamic range of speech sounds-30 dB
i.e. the difference between the softest and most ________
III. Ways to Represent Speech
________
________
________
Waveform
________
Spectrum
________
Spectrogram
________
IV. Effects of Hearing Loss
A. Reduction in Absolute Sensitivity
Speech sounds are not audible
B. Reduction in Differential Sensitivity
Speech sounds cannot be distinguished in time, intensity or frequency
A.Reduction in Absolute Sensitivity
Speech errors depend on the frequency region and the severity of the hearing loss
Frequency Region of Speech Cues
Nasality-Low Frequency
Frication, Sibilance- High Frequency
Voicing, Manner-Low to High Frequency
Place-High Frequency
Examples: To hear these
differences, one needs hearing in the ______
frequency region.
Sheet vs Heat - High
Man vs Pan - Low
Tea vs Key - High
Door vs Tore - Low to Mid
B. Reduction in Differential Sensitivity
Speech errors depend on ability to resolve changes in time, intensity or frequency
1) Temporal Resolution-Can tell the difference between 20 and 10 ms gap in noise
2) Intensity Resolution-Can recognize a 2 dB increase in intensity
3) Frequency Resolution-Can tell the difference between energy at 2000 vs 2200 Hz
Examples: To hear these differences, one needs resolution in _________ domain.Cot vs Got
Time-To hear Voice Onset Time
Buy vs Die vs Guy
Frequency-To hear F2 transitions
Bad girl vs Sad girl
Intensity-To hear stress on "Bad"
Over connected speech, reduced differential sensitivity can result in difficulty recognizing word boundaries as timing information may be smeared.
That’s one reason why persons with hearing loss are at such a disadvantage when listening in background noise.
V. Evaluation of Speech Recog
A. Evaluation of Absolute Sensitivity
1. Involves presenting words or sentences & patient points to picture or repeats words.
2. Usually get % correct score.
3. Problem is there is a high degree of unreliability unless you use 100 words or more.
Variability illustrated by Thornton & Raffin
If score 80% on a list of 50 words then on a subsequent test you'd have to score above 92% or below 64% in order to say significant change had occurred.
OR if score 80% on a list of 25 words then critical difference range is ____ to ____.
A sensitive tool to quantify
audibility of the speech signal is the
Articulation Index
The AI varies from 0 to 1 and
represents the proportion of the speech signal that is audible.
It can be calculated from thresholds and uncomfortable loudness levels.
ex. A hearing aid that resulted in an AI of .8 should be better than one that results in an AI .6
Remember.....The AI just reflects absolute sensitivity not differential sensitivity. There is some debate that we should only be concerned with absolute sensitivity.
i.e...........
In fitting a hearing aid all we need to be concerned with is making sounds audible.
The less severe the hearing loss, the more likely this is true.
B. Evaluation of Differential Sensitivity
1. This is not routinely done in the clinic
2. There are adaptations of psychophysical paradigms that have been proposed
3. The problem is that the relationship between frequency or temporal resolution and speech recognition has not been firmly established
Temporal resolution
Hearing impaired listeners (sensori-neural) with moderate - profound losses tend to have abnormal temporal resolution
1) ie- they need bigger gaps to tell that a temporal change occurred.
2) Recall....this may result in difficulty with voicing cue (voice onset time)
Frequency resolution
Hearing impaired listeners (sensori-neural) with moderate - profound losses tend to have abnormal frequency resolution
1) ie-they need a greater change in frequency to tell that a change has occurred.
2) Recall....this may result in difficulty with place information (second formant transitions)
One way to quantify deficit in differential sensitivity is to see how HI performance differs from normals under similar conditions.....ie give normals speech recognition test with filtered words or added background noise and compare HI performance