Lecture 4- Speech Threshold Assessment Flashcards
What is the speech stimulus? How can the speech stimulus be represented on a speech spectrum?
Speech stimulus: a complex signal that varies in amplitude, duration, and frequency components
- On the spectrum, speech is composed of many frequencies
Peak: 500-700 Hz
Rolloff above 1000 Hz: 10-12 dB/octave - There are similar spectrums for males, females, and different speakers
Relative intensity of speech components:
- Strongest components: vowels
- Weakest components: consonants
What is the LTASS?
Long-term average speech spectrum
- Most of the energy in speech is in the low frequencies due to the fundamental frequency from the voice and vowels
What is speech detectability?
Faintest level at which an individual can detect speech 50% of the time (SDT- speech detection threshold)
- Also known as the speech awareness threshold*
What is speech intelligibility?
Lowest intensity where the individual correctly repeats the material 50% of the time (SRT- speech reception threshold)
What is absolute intelligibility?
The faintest level at which the individual repeats 100% of the material
What are the relationships between the different speech thresholds?
- SDT is 8-9 dB < SRT, which is 4-5 dB less than the threshold of absolute intelligibility
- SDT = best threshold on audiogram (250-4000 Hz)
- SRT = average hearing in speech frequencies (500-2000 Hz)
Why do audiologists use speech thresholds?
1) To validate the pure tone audiogram
- Evaluate the relationship between sensitivity for speech and pure tones
2) Use as a reference intensity level
- Supra threshold speech recognition (dB SPL re: SRT)
3) Used to estimate hearing sensitivity
- For difficult to test patients
4) Used to determine hearing aid candidacy, estimate of benefits, monitor performance
What are some historical speech threshold tests?
1) Whispered speech tests
- Could occur in any doctor’s office
2) Invention of phonograph (1877) - Edison
- Could present recorded material
- Intensity could be controlled
3) First speech audiometer (1926) - Bell Labs
- Presented phonographic recordings (digit pairs)
- Reduced intensity in 3-dB steps
- Measured articulation function for speech (% correct vs. presentation level)
- SRT corresponds to 50% point
- Developed Bell Telephone Intelligibility Lists (49 lists; 50 sentences)
4) Monitored Live voice testing (1942)
- Hughson & Thompson
- Hooked v-u meter to speech audiometer to monitor intensity of voice
- Measured SRT via pure-tone audiometric technique
- Noted a linear relationship between pure tone & speech threshold
5) CID (1950s) Hirsh - spondee disc recordings
- Stimuli: more familiar words than PAL
- Recordings: W1, constant level, W2, stair step
Tabular methods: start at suprathreshold level
- 3 words per step, in 3 dB steps
- Assign 1 dB for each correct word; count # of words correct
- Subtract # correct from starting level, add correction factor
Efforts to produce “homogeneous” list of spondees
- Threshold for different spondees obtained at similar levels
- Lists available with 15 selected spondees
What did the Harvard Psychoacoustics Labs (PAL) create?
Hudgins, et. al., 1947
- Developed materials for measuring SRT
- Criteria for selection of words used in test:
1- Highly audible and equally audible
2- Familiar vocabulary
3- Phonetic dissimilarity
84 words chosen = all spondees (= stress on 2 syllables)
- Recordings of spondee word lists
- Developed tabular form for measuring SRT
Describe the Tillman-Olsen Method for obtaining thresholds.
- A descending method with 2 phases
Preliminary phase
- Read words at 5 dB at supra threshold level, w/o visual cues
- Discard any words that the listener can’t repeat
- Give the listener the instructions
- Present the words formally at supra threshold level
- Present 1 word, if correct, descend by 10 dB
- Continue until listener misses 2 words @ 1 level
- Increase level by 10 dB (this is the starting level)
Test Phase - record responses right or wrong
- Present 2 words per level
- Subject must repeat 1st 5/6 words correctly
- Descend in 2 dB steps
- Continue until listener misses last 5/6 words
Calculation of threshold?
T = starting level - # correct + 1 dB
What are the advantages & modifications of the Tillman- Olsen Method?
Advantages
- Standardized procedure
- Removes tested bias
- Highly reliable
Modifications
- Can use 5 dB steps, with 5 words on a step
- Correction factor = 2 dB
What is the current method for obtaining the SRT?
- Jahner, Schlauch, & Doyle (1994)
- Compared ASHA 1979 and T-O Methods for measuring SRT
Results:
- SRT with T-O method < 1979 method by 4 dB
- ASHA 1979 method agreed better with PTA and took less time to perform
Conclusions: for clinical purposes, two methods are similiar
What are variables that influence measurement of SRT?
- Descending vs. ascending method (not significant)
- Step size, 2 dB or 5 dB (not significant)
- Carrier phrase vs. no carrier phrase ( not significant)
Recorded vs. Monitored Live Voice
- MLV: widely used (flexible for kids/elderly)
- MLV: difficult to peak at 0 BU
- Results are reliable if they are the same tester
- To use standardized procedure- must use recorded material
Effect of prior knowledge of vocabulary and practice effects
- Improvement in SRT due to practice: 1 dB
- Improvement in SRT due to practice and possible prior knowledge (2.4 dB)
- Improvement in SRT due to prior knowledge of vocabulary (6 dB)
Conclusion: practice doesn’t influence thresholds too much, but prior knowledge does
Recommend: familiarization of vocabulary
What is the relationship between speech and pure tone thresholds?
Comparison using SPL scale
Ex. SRT = 20 dB SPL
PTA = 11.5 + 7 + 9 dB SPL
= 9 dB SPL
Comparison using HL scale:
- Threshold for speech is approximately equal to the thresholds for pure tones
Acceptable agreement between PTA and SRT: +/- 6 dB –> +/- 12 dB
Reasons for lack of agreement
SRT
What are the methods for calculating pure tone average (PTA)?
Fletcher (1929) - 3 frequency PTA
-Average threshold for 500, 1000, and 2000 Hz, used for predictor of speech threshold
Carhart (1946)
- Correlation between PTA & SRT depends on audiogram configuration
- r = .79 (flat), r = .75 (gradually sloping), r = .29 (sharply sloping)
- Better SRT than PTA because speech is concentrated in low frequencies
Fletcher (1950): use average of best 2 thresholds in 500 to 2000 Hz range
- Works best when applied to sloping audiogram
- Use PTA or Fletcher average depending on slope of audiogram