What Is a Speech Recognition Test?

What question does the test answer?

Word-recognition testing asks: **When speech is made audible, how accurately can this ear identify the words presented under these specific conditions?**

It can help an audiologist:

Compare speech understanding between the ears

Determine whether louder speech becomes clearer

Add context to the pure-tone audiogram

Counsel about communication expectations

Support hearing-aid or implant evaluations

Decide whether additional testing or medical referral is appropriate

Establish a baseline for future comparison

The score is not a complete measure of everyday communication. It reflects performance on one set of materials, at one level, in a particular test environment.

Speech-recognition threshold versus word recognition

These two tests are often confused because both involve spoken words.

| Speech-recognition threshold | Word-recognition testing |
|---|---|
| Often abbreviated SRT | Sometimes abbreviated WRS or word recognition |
| Uses familiar words near threshold | Uses a list of words above threshold |
| Finds the softest level for about 50% recognition | Calculates a percentage correct |
| Reported in dB HL | Reported as a percentage plus test conditions |
| Helps confirm agreement with tone thresholds | Examines clarity when speech is audible |

When reviewing a hearing report, ask whether “speech recognition” refers to the SRT, the word-recognition percentage or both.

How is word recognition tested?

Preparing the test ear

Each ear is usually tested separately. The audiologist reviews the pure-tone results and chooses an appropriate speech presentation level.

Speech must be audible enough to test recognition, but the level must also remain safe and comfortable. The selected level may differ between ears.

Listening to the word list

The person hears a series of single words, often without sentence context. After each word, the person repeats it exactly or selects a matching response.

Using isolated words reduces the ability to guess from context. The task may feel more difficult than ordinary conversation, where grammar, topic and facial expressions provide clues.

Scoring responses

Each correctly identified word contributes to the final percentage. If 25 words are scored, each word represents four percentage points. If 50 words are scored, each word represents two points.

The list length matters because a short list creates larger percentage steps and greater uncertainty. A small difference between two scores may reflect normal measurement variability rather than a real change.

Testing the other ear

The procedure is repeated for the opposite ear. If one ear hears much better, masking noise may be needed to prevent it from responding to speech intended for the poorer ear.

What do the words sound like?

Many clinical word-recognition lists use familiar one-syllable words. They may be presented from a calibrated recording or spoken live by the clinician.

Recorded materials provide more consistent pronunciation, timing and level from one test to another. Live-voice testing allows flexibility but introduces variation in the talker’s voice and delivery.

The report should ideally identify the test material and whether it was recorded or live, especially when results are compared across visits.

What does the percentage mean?

A word-recognition percentage shows the proportion of scored words identified correctly under the test conditions.

For example:

24 correct responses out of 25 words equals 96%.

20 correct responses out of 25 words equals 80%.

15 correct responses out of 25 words equals 60%.

The percentage should never be interpreted alone. The audiologist also considers:

Presentation level

List length

Word list and language

Recorded or live presentation

Right- versus left-ear difference

Masking

Pure-tone thresholds

Reliability and consistency

Age, attention and communication abilities

Whether testing occurred in quiet or noise

A clinic may use descriptive categories, but the boundaries are not universal. The most useful explanation connects the score to the full hearing profile and the person’s communication needs.

Why presentation level matters

Speech recognition can change depending on how loudly the words are presented.

If the level is too soft, parts of the words may be inaudible and the score may underestimate potential performance. If the level is uncomfortably loud, performance or tolerance may also be affected.

The audiologist may select a level relative to the SRT, hearing thresholds or comfort range. In some cases, testing at more than one level can help identify the best performance or examine whether recognition changes with intensity.

Therefore, “I scored 72%” is incomplete without knowing **at what level** and under **which protocol** the test was performed.

Why louder does not always mean clearer

Hearing loss can reduce audibility, clarity or both. Amplifying speech can restore access to some sounds, but it cannot always correct distortion caused by inner-ear or neural damage.

A person may hear that a word was spoken yet confuse similar speech sounds. Increasing the level may improve performance up to a point, then produce little additional benefit.

In certain patterns, the audiologist may examine whether performance becomes poorer at higher levels. Such a finding does not diagnose a specific disorder by itself, but it may contribute to a decision for additional evaluation.

Why the ears may have different scores

One ear can understand words less accurately than the other, even when the pure-tone thresholds look similar.

Possible contributors include differences in hearing sensitivity, speech audibility, inner-ear function, auditory-nerve function, presentation level, test reliability or language factors.

A significant or unexpected asymmetry may prompt retesting, additional audiologic measures or medical referral. The percentage alone does not identify the cause.

What is masking?

Speech presented at a sufficiently high level to one ear may cross through the skull and be detected by the opposite ear. This can make the poorer ear appear to perform better than it actually does.

To obtain an ear-specific result, the audiologist may present controlled noise to the non-test ear. This is called **masking**.

The noise may sound like static or rushing air. It is a normal part of testing when the difference between ears or the presentation level makes cross-hearing possible.

Why list length matters

A word-recognition score is based on a sample, not every word a person might hear. Short lists are faster, but each response has a larger effect on the final percentage.

For a 25-word list, one changed response moves the score by four points. For a 50-word list, one response changes it by two points.

Scores also have statistical variability. When comparing visits or ears, the audiologist considers list length and whether the difference is large enough to be meaningful. Two percentages should not automatically be treated as different simply because the numbers are not identical.

Why language and vocabulary matter

Word recognition requires familiarity with the test language and vocabulary. A person may hear the acoustic signal accurately but respond incorrectly because the word, accent or response requirement is unfamiliar.

Results may be influenced by:

Primary language

Multilingual experience

Accent or dialect

Vocabulary

Literacy

Speech-production ability

Cognitive or developmental factors

Attention and fatigue

Tell the clinic which languages and communication methods are used. The audiologist may select different materials, use picture choices, measure speech detection or document limitations in interpretation.

A score from an inappropriate word list should not be treated as a pure measure of auditory function.

Word recognition in quiet versus speech in noise

Traditional word-recognition testing is often performed in quiet. This provides a controlled ear-specific measure but may not represent the environments where a person struggles most.

Speech-in-noise tests add competing sound and may use words, digits or sentences. They can help evaluate complaints such as difficulty in restaurants, meetings or group conversations.

| Words in quiet | Speech in noise |
|---|---|
| Controlled single-word task | Competing sound is present |
| Usually reported as percentage correct | May report percentage or signal-to-noise result |
| Useful for comparing ears | Often closer to common real-world complaints |
| Limited contextual cues | Difficulty depends on the specific noise and protocol |

A high score in quiet does not guarantee easy understanding in noise.

Does a high score mean hearing is normal?

No. A high word-recognition score means the person performed well on that list at that presentation level.

Someone can have hearing loss on the audiogram and still achieve excellent recognition when the words are made loud enough. They may continue to miss soft speech, distant voices or conversation in background noise.

Pure-tone thresholds, speech-in-noise results, symptoms and everyday experience must also be considered.

Does a low score diagnose the cause?

No. A reduced score can result from several factors, including insufficient audibility, inner-ear distortion, neural involvement, language mismatch, presentation level, fatigue or inconsistent responses.

The audiologist interprets the score with the audiogram, case history, middle-ear findings and other tests. Unexpected or asymmetric performance may justify additional diagnostic or medical evaluation.

How results may affect hearing-aid recommendations

Word-recognition results can help set expectations, but they do not predict hearing-aid success perfectly.

Hearing aids may improve access to speech by making important cues audible. They cannot guarantee that every amplified word will be understood, especially when unaided recognition remains limited or background noise is strong.

Good hearing-aid care also includes:

Needs assessment

Individual device selection

Real-ear or other objective verification

Aided speech testing when appropriate

Communication counseling

Follow-up adjustments

Outcome measurement in daily life

Real-ear verification checks the sound delivered by the hearing aid in the ear. Word-recognition testing measures performance on a speech task. They answer different questions.

Speech recognition and cochlear implant evaluation

Implant candidacy is not determined by an unaided word-recognition percentage alone. A cochlear implant evaluation includes specialized aided speech testing with appropriately fitted hearing aids, medical assessment and review of hearing history and goals.

Referral for an evaluation may be appropriate when hearing aids provide limited speech understanding or insufficient functional benefit. An evaluation does not obligate someone to proceed with surgery.

Speech recognition testing for children

Children need materials and responses matched to their developmental age, language and vocabulary.

The audiologist may use:

Picture-pointing tasks

Familiar object or body-part identification

Age-appropriate word lists

Speech detection when word identification is not yet possible

Aided testing with hearing technology

Results are interpreted with pure-tone or behavioral thresholds, objective measures, developmental history and caregiver observations.

What if someone cannot repeat words?

Spoken repetition is not the only possible response. Depending on the goal, a person may point to pictures, select written choices, type responses or indicate only that speech was heard.

Motor, speech, cognitive and language factors should be accommodated. If a standard percentage cannot be measured validly, the audiologist documents the alternative method and its limitations.

How reliable is the test?

Reliability improves when:

Instructions are clear

The person understands the vocabulary

Speech is presented at an appropriate level

Equipment and materials are calibrated

Masking is used when necessary

The list is long enough for the clinical purpose

Responses are consistent

Fatigue and attention are considered

If a result does not fit the rest of the evaluation, the audiologist may repeat the test, change the presentation level, use a different list or add other measures.

Can an online speech recognition test replace clinical testing?

No. An online word or digits test may screen for possible difficulty, but home equipment and headphones are not generally calibrated like clinical systems. Background noise, device settings, language and test design can change the result.

Online tests usually cannot perform appropriate ear-specific masking, examine the ears, test bone conduction or assess middle-ear function. They cannot diagnose the type or cause of hearing loss.

Do not use online testing for sudden hearing change, trauma, severe dizziness, neurological symptoms, ear pain, bleeding or drainage. Seek professional care.

How to prepare

Bring previous audiograms and speech scores if available.

Bring current hearing aids and accessories.

Tell the audiologist which languages and communication methods you use.

Report tinnitus, dizziness, pain, pressure or drainage.

Share relevant medical, medication and noise history.

Request an interpreter or other accommodation in advance.

During the test, repeat exactly what you heard—even if the response sounds incorrect or is not a real word. Do not wait for certainty; the clinician expects some items to be difficult.

Questions to ask your audiologist

Was this an SRT or a word-recognition test?

Which word list was used?

Was the material recorded or spoken live?

How many words were scored?

At what level were the words presented?

Was masking used?

How certain is the score given the list length?

Are the right and left ears meaningfully different?

How does the score compare with my audiogram?

Would speech-in-noise or aided testing add useful information?

Do the results suggest medical or implant evaluation?

The bottom line

A speech recognition test can show how accurately spoken words are identified under controlled conditions. The percentage is useful only when interpreted with the presentation level, word list, list length, language, masking and other hearing results.

A high score does not prove hearing is normal, and a low score does not diagnose a specific condition. The audiologist combines the result with pure-tone testing, symptoms and real-world communication needs. Sudden change in hearing or speech understanding requires immediate medical evaluation.

Use the **Audiology.md provider directory** or **Request Hearing Help** to connect with local hearing care near you.