Skip to content

LESSON 95

Speech recognition threshold: SRT

The speech recognition threshold (SRT) is the lowest level at which fifty percent of two-syllable words are repeated correctly, and it is used to cross-check pure tone thresholds. In this lesson we cover spondee words, familiarization, the ASHA descending method, a sample SRT calculation and agreement between the SRT and the pure tone average (PTA).

Watch on YouTube

Follow along

Hello, and welcome to the AIHEAR Audiology Academy.
In this lesson we will learn, step by step, how to find the speech recognition threshold, or SRT.
We will look at word selection, the familiarization phase and how the result is compared with the pure tone average.
The speech recognition threshold, or SRT, is the basic threshold measure of speech audiometry.
The SRT is the lowest intensity level at which a person can correctly repeat fifty percent of two-syllable words.
In older sources the same measure is also called the speech reception threshold.
Like pure tone thresholds, the result is recorded in decibels HL, separately for each ear.
The most important role of the SRT is to check the reliability of the pure tone thresholds.
It is also used to decide the level at which the word recognition test will be given.
The SRT uses spondees, which are two-syllable words with equal stress.
In these words both syllables are spoken with similar stress and intensity.
The words in a list are chosen so that they all start to be recognized at about the same intensity level.
Thanks to this, the percentage of correct responses changes quickly with small level changes, and the threshold is clear.
Because Turkish has a different stress pattern, common Turkish lists use three-syllable words chosen for equal intelligibility.
We will cover Turkish speech test materials in detail in a separate lesson.
Speech tests are given through the speech channel of the audiometer.
Live voice through a microphone, or a recorded source, is connected to this channel.
The words are presented to the patient through earphones, to each ear separately.
The audiologist hears the patient's spoken response through the talkback microphone in the booth.
So for the SRT, the patient repeats the word instead of pressing the response button.
The calibration of the speech channel must be checked regularly, just like the pure tone channel.
Before testing begins, there is a familiarization phase.
In this phase the word list is introduced to the patient, spoken or written, at a comfortable listening level.
Words the patient does not know or cannot repeat correctly are removed from the list.
Familiarization reduces errors caused by vocabulary and makes the threshold more reliable.
This is why the guideline of the American Speech-Language-Hearing Association, or ASHA, also recommends familiarization.
The instructions must be clear too: the patient is asked to repeat the word they hear, and to guess if unsure.
For patients who cannot speak, or for young children, a picture-pointing response can be used instead of repetition.
Words can be presented from recorded material or read by the audiologist into the microphone, that is, by live voice.
Recorded material is more standardized, while live voice is more flexible and faster.
With live voice, the level of each word is kept as steady as possible by watching the VU meter on the audiometer.
The speaker also takes care to say the words naturally and with the same intonation.
The test is done with earphones, separately for each ear.
If the presentation level is high enough to cross to the opposite ear, that ear is masked with speech noise.
In the descending method recommended by ASHA, the first word is given about thirty to forty decibels above the estimated threshold.
As long as the patient repeats correctly, the level is lowered in ten decibel steps.
When two words in a row are missed at one level, ten decibels above that level becomes the starting level of the test phase.
In the test phase, five words are given at each five decibel step and the number correct is recorded.
The test stops when the patient misses five of the last six words.
There is also a version of this method that uses two decibel steps, with two words at each step.
Let us work through an example: the test phase starts at forty decibels, and the patient repeats all five words correctly.
At thirty five decibels, again five out of five are correct.
At thirty decibels, four of the five words are correct.
At twenty five decibels there are only two correct responses.
At twenty decibels there are none; at least five of the last six words are wrong, so the test stops here.
The calculation: the starting level is forty, the total correct is sixteen, and the correction for five decibel steps is two.
The SRT is forty minus sixteen plus two, which is twenty six dB HL.
The general formula is: subtract the total number correct from the starting level, then add the correction factor.
The correction factor is two for five decibel steps and one for two decibel steps.
This calculation is a statistical estimate of the fifty percent correct point.
Some clinics also use ascending methods, similar to pure tone threshold searching.
Whichever method is chosen, it is important to apply it consistently within the same clinic.
The SRT is compared with the average of the pure tone thresholds, called the pure tone average, or PTA.
The PTA is usually the average of the thresholds at five hundred, one thousand and two thousand hertz.
The difference between the two values usually stays within plus or minus ten decibels, and in most patients it is even smaller.
In this example, the right ear PTA is about twenty seven and the SRT twenty six dB HL; the two values agree.
Agreement shows that the pure tone thresholds and the speech threshold confirm each other.
The speech detection threshold, or SDT, is usually about eight to ten decibels better than the SRT.
If the SRT is clearly better than the PTA, an important possibility to consider is functional, or nonorganic, hearing loss.
These people may show raised thresholds for pure tones, yet repeat speech at a much lower level than expected.
With a steeply sloping audiogram, the SRT can also come out better than the three-frequency PTA.
In this case the SRT agrees better with the average of the best two of these three thresholds; this is called the Fletcher average.
If the SRT is clearly poorer than the PTA, a language difference, cognitive difficulty or a speech understanding problem may be considered.
In either direction, first check that the instructions were understood and that the equipment is working correctly.
Here is a habit that is very useful in the clinic.
Compare the SRT result with the PTA while the patient is still in the booth.
If the difference is more than ten decibels, repeat the instructions and recheck the pure tone thresholds before writing the report.
Many reliability problems are caught by this simple comparison before the test is over.
When functional loss is suspected, some clinicians prefer to measure the SRT before pure tone testing.
The first use of the SRT is to confirm the pure tone thresholds.
The second is to set the presentation level of the word recognition test, often thirty to forty decibels above the SRT.
In some patients who cannot respond reliably to pure tones, the speech threshold can give an idea of hearing level.
For example, in young children, a speech threshold found with a picture-pointing response is valuable information.
A speech threshold measured in sound field can also be used to compare aided and unaided hearing.
The SRT is the lowest level at which fifty percent of two-syllable words are repeated correctly.
The test uses equally intelligible two-syllable words, after a familiarization phase.
In the ASHA method, the SRT is the starting level minus the number correct, plus a correction factor.
The SRT and PTA usually agree within ten decibels; a large difference calls for rechecking the result.
Test yourself: the starting level is fifty decibels, the total number correct is eighteen, and the step is five decibels. What is the SRT?
The correct answer is B: thirty four dB HL.
Fifty minus eighteen plus two is thirty four; with five decibel steps the correction factor is two.
In the next lesson we will learn how to measure the word recognition score, a suprathreshold test.
Subscribe to the channel so you never miss a new lesson. See you next time!

Keywords

Click a keyword to search all lessons for that term.

Prerequisites

LESSON 94 Why Speech Audiometry? Goals and Test Types

Comments

Questions and comments appear after approval.

No comments yet. Be the first to comment.