Speech Audiometry Explained

What does speech audiometry measure?

Depending on the test, speech audiometry may measure:

The softest level at which speech is detected

The softest level at which familiar words are recognized

How accurately words are repeated at a comfortable listening level

How understanding changes as speech becomes louder

How well sentences or words are understood in background noise

Performance with hearing aids or cochlear implants

Differences between the right and left ears

The audiologist chooses tests based on the reason for the visit, age, language, hearing thresholds and ability to respond.

Speech testing versus pure-tone testing

Pure-tone audiometry and speech audiometry measure related but different abilities.

| Pure-tone audiometry | Speech audiometry |
|---|---|
| Uses simple tones | Uses words, sentences or speech-like signals |
| Measures detection by frequency | Measures detection or understanding of speech |
| Produces an audiogram | Produces thresholds, percentages or signal-to-noise results |
| Helps classify degree and type | Helps describe functional speech access |
| Does not directly measure word understanding | Tests speech recognition under defined conditions |

The results should be interpreted together. Agreement between certain tone and speech thresholds can support test reliability, while unexpected differences may lead to reinstruction or additional assessment.

The main types of speech audiometry

Speech detection threshold

The **speech detection threshold**, or SDT, is the lowest level at which a person can detect that speech is present about half the time. It may also be called the speech awareness threshold, or SAT.

The person does not need to identify the words. They only indicate that speech was heard. This can be useful for young children, people who do not speak the test language or anyone who cannot reliably repeat words.

An SDT measures awareness, not understanding.

Speech-recognition threshold

The **speech-recognition threshold**, or SRT, is the lowest level at which a person can correctly recognize approximately half of a set of familiar words. Traditional testing often uses two-syllable words with roughly equal stress on each syllable, sometimes called spondees.

The SRT is measured in dB HL and is often compared with the pure-tone average across important speech frequencies. Reasonable agreement can help confirm the consistency of the behavioral results.

Despite the similar abbreviation, SRT is not the same as a word-recognition percentage measured at a louder level.

Word-recognition testing

Word-recognition testing measures how accurately a person repeats a list of single words presented at a selected level above threshold. It may be called word recognition, speech discrimination or suprathreshold speech-recognition testing.

The result is usually reported as a percentage correct for each ear. For example, correctly repeating 20 of 25 scored words would produce an 80% score for that list.

The score must be interpreted with the presentation level, test material, list length, language and test conditions. A percentage without those details can be misleading.

Speech-in-noise testing

Speech-in-noise testing measures understanding when competing sound is present. Depending on the test, the result may be a percentage correct or a signal-to-noise value showing how much louder speech must be than background noise for a target level of performance.

This testing can reflect a common real-life complaint more closely than single words presented in quiet. It may also guide counseling, remote-microphone recommendations and hearing-technology settings.

What happens during speech testing?

The person usually sits in a quiet or sound-treated space and listens through insert earphones, headphones, speakers or hearing devices.

The audiologist may ask the person to:

Raise a hand when speech is detected

Repeat familiar words

Repeat single words exactly as heard

Repeat sentences in quiet or noise

Select a picture that matches a word

Enter digits on a screen or keypad

Respond through another accessible method

Speech may be played from a recording or presented live by the clinician. Recorded materials provide more consistent presentation. Live-voice testing may be used when flexibility is needed, but the talker’s voice can influence results.

Understanding the speech-recognition threshold

The SRT answers a limited question: how soft can familiar speech be before recognition becomes unreliable?

It does not show how well someone understands normal conversation, rapid speech or speech in noise. It also does not measure the best possible word-recognition score.

The SRT is useful because it can:

Confirm general agreement with pure-tone thresholds

Help select levels for other speech tests

Provide a basic speech audibility measure

Support testing when tone responses are uncertain

If the SRT and pure-tone average differ more than expected, the audiologist may review instructions, equipment, language factors and test reliability.

Understanding word-recognition scores

Word-recognition testing is performed above threshold so the words are audible. The audiologist selects a presentation level based on hearing thresholds, comfort, clinical goals and test protocol.

A lower score can suggest that making sound louder does not provide full clarity. However, the result is not interpreted from the percentage alone.

Important factors include:

Which word list was used

How many words were presented

Whether speech was recorded or live

The presentation level

Whether the level was comfortable

Whether masking was used

The person’s language and familiarity with the words

Attention, fatigue and response consistency

Whether the test was performed in quiet or noise

Different list lengths also create different statistical uncertainty. A change of several percentage points may not represent a meaningful change, especially with a short list.

Why louder speech may not become clearer

Hearing loss can reduce audibility, clarity or both. When speech is too soft, amplification may restore access to some cues. When the inner ear or auditory nerve cannot transmit details accurately, added volume may not restore full word understanding.

Some people reach a best score and then remain stable as the level increases. In certain hearing conditions, performance may decline at higher presentation levels. The audiologist decides whether testing across multiple levels is appropriate.

“I can hear you, but I cannot understand you” is therefore a valid and clinically important report.

What is masking in speech audiometry?

Speech presented loudly to one ear may sometimes be detected by the opposite ear. This is especially important when one ear hears much better than the other.

The audiologist may present controlled noise to the non-test ear to keep it from helping. This process is called **masking**.

Masking noise may sound like static or rushing sound. It is a standard method for obtaining ear-specific results. It does not mean the hearing loss became worse or the equipment malfunctioned.

Speech testing in quiet versus noise

Testing in quiet

Single words in quiet create a controlled measure of recognition for each ear. They are useful for comparing ears and examining performance when audibility is provided.

However, quiet single-word testing does not reproduce a restaurant, family gathering or meeting. It may overestimate how easy everyday communication feels.

Testing in noise

Speech-in-noise tests add competing talkers, steady noise or another controlled background. The task may use sentences, words or digits.

The results can help explain why someone performs well in quiet but struggles in complex environments. They may support recommendations such as directional microphones, remote microphones, seating changes, captions or communication strategies.

Speech-in-noise tests are not interchangeable. A score should be interpreted using the specific test’s procedure and reference values.

Signal-to-noise ratio explained

The **signal-to-noise ratio**, or SNR, compares the level of the target speech with the background noise.

A positive SNR means speech is louder than the noise.

A 0 dB SNR means speech and noise are at the same level.

A negative SNR means speech is softer than the noise.

Some tests estimate the SNR a person needs to achieve a defined level of sentence or word understanding. Needing a more favorable SNR can indicate greater difficulty in noise.

The number does not mean every real-world environment will produce the same performance. Distance, reverberation, talker differences, visual cues and fatigue also matter.

Why the right and left ears are tested separately

Ear-specific testing can identify asymmetry that a sound-field or both-ears-together result may hide.

A significant difference in word recognition or hearing thresholds may affect treatment recommendations and, depending on the pattern and symptoms, may prompt medical referral.

Testing each ear also helps determine whether two hearing aids, a CROS system, implant evaluation or another strategy should be considered. The complete evaluation—not one score—guides the decision.

Aided speech testing

Aided testing measures performance while a person uses hearing aids or a cochlear implant. Speech may be presented through speakers in quiet or noise.

It can help evaluate:

Whether speech is audible with the devices

Functional benefit compared with unaided performance

Performance in noise

Whether device adjustments may be needed

Whether implant evaluation criteria should be explored

Aided speech testing does not replace real-ear verification of a hearing-aid fitting. Verification confirms the device output in the ear; aided testing examines performance on a listening task. Both can be useful.

How speech results guide hearing care

Speech audiometry may influence:

Hearing-aid counseling and expectations

Device selection and programming

Remote-microphone or assistive-listening recommendations

Communication-strategy training

Aural rehabilitation

Cochlear implant referral

Medical referral when results are asymmetric or unexpected

The need for additional diagnostic testing

The audiologist should connect the test results to the person’s specific listening goals rather than treating one score as the outcome.

Language, accent and cultural considerations

Speech testing depends on the listener’s familiarity with the language and test vocabulary. Accent, dialect, multilingual experience, literacy, articulation and cultural background can affect performance.

A low score may not represent auditory ability accurately if the material is unfamiliar or the response method is inappropriate.

Tell the clinic which languages and communication methods are used. The audiologist may choose alternative materials, an interpreter, picture-pointing tasks, speech detection or objective measures. Interpretation must acknowledge the limits of the selected test.

Speech audiometry for children

Children require developmentally appropriate words, pictures and response methods. The audiologist considers age, vocabulary, language exposure, attention and communication mode.

Possible approaches include:

Speech detection for infants or children not ready to identify words

Picture identification

Body-part or object identification

Familiar word repetition

Age-appropriate speech-in-noise tasks

Aided testing with hearing devices

Speech results should be integrated with behavioral tone testing, objective measures, developmental history and caregiver observations.

Speech audiometry for people who cannot repeat words

Repeating aloud is not the only valid response. Depending on the clinical question, a person may point to pictures, select written choices, type a response or indicate only that speech was detected.

Motor, speech, cognitive or language limitations must be considered. The audiologist adapts the test and clearly documents what was measured.

An SDT may be useful when word repetition is not possible, but it does not provide a word-understanding score.

Can normal word recognition rule out a hearing problem?

No. A good word-recognition score in quiet means the person performed well on that particular list at that presentation level. It does not guarantee normal hearing, easy conversation in noise or normal auditory processing.

Pure-tone thresholds, speech-in-noise performance, middle-ear findings, symptoms and real-world difficulties still matter.

Can a low score diagnose the cause?

No. A low score may be associated with reduced audibility, inner-ear distortion, neural involvement, language mismatch, presentation level, fatigue or other factors.

The audiologist compares the result with the full test battery and history. Medical or specialized diagnostic evaluation may be recommended when findings are unexpected or significantly asymmetric.

Can an online speech test replace clinical testing?

Online speech or digits-in-noise screening may identify possible difficulty, but it does not replace a comprehensive evaluation.

Results can be affected by headphone quality, calibration, background noise, language, internet or device behavior and test design. Most online tests cannot examine the ears, perform bone conduction, assess middle-ear function or provide a complete ear-specific diagnostic battery.

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 hearing reports if available.

Bring current hearing aids, chargers and accessories.

List the listening situations that are most difficult.

Tell the audiologist which languages and communication methods you use.

Report tinnitus, dizziness, pain, pressure or drainage.

Share relevant medical, medication and noise-exposure history.

Request an interpreter, captions or another accommodation in advance.

During testing, repeat what you heard—even if the word does not make sense. Avoid filling in a different word based on context unless instructed otherwise.

Questions to ask your audiologist

Which speech tests were performed, and what did each measure?

What was my speech-recognition threshold in each ear?

At what level was word recognition tested?

Were the words recorded or presented live?

Was masking used?

How reliable is the percentage given the list length?

How do my speech results compare with my audiogram?

Would speech-in-noise testing add useful information?

Do the ears differ enough to require medical referral?

How do these results affect my hearing-technology options?

The bottom line

Speech audiometry measures several different aspects of speech access. Speech detection and recognition thresholds show how soft speech can be detected or identified. Word-recognition testing examines accuracy at an audible level, while speech-in-noise testing evaluates performance with competing sound.

No single score describes everyday communication or establishes a medical diagnosis. Results are most useful when interpreted alongside pure-tone thresholds, ear health, history and real-world listening needs. Sudden hearing or speech-understanding change requires immediate medical evaluation.

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