Amp week 6B Speech Acoustics

The Core Goal of HAs is to

Improve Speech Audibility

Amplitude Modulations

• Modulation rate

• Range from 30 Hz (e.g., 33 ms) for the shortest phoneme

• 0.5 Hz (e.g., 2 sec) for some sentences

Modulation depth

how large of a change in amplitude there is from the maxima of the signal to its minima.

Speech Envelope

Used by HA algorithms to identify and separate speech from noise

Effect of Reverberation

fills in the modulation depths within speech

Signal-to-noise ratio (SNR)

level of speech in comparison with the level

of noise

• + SNR = speech is higher than the noise

• - SNR = noise is higher than the speech

Individuals with NH require a SNR of

+6 dB for satisfactory communication

Individuals with SNHL require

a significantly greater SNR to achieve similar performance with individuals with NH

Long-Term Average Speech Spectrum (LTASS)

• Graphic plot of level across frequency (speech spectrum) after averaging over some predefined segment of time

• Defined by Vocal Effort for Production

• Soft, Conversational, Loud, Shouted

Direct measure of the mean SPL of speech as a function of frequency and provides a global representation of the acoustic characteristics of continuous discourse

LTASS Variables:

• Gender

• Language

• Vocal Effort

• Vocal Hygiene

• Speech material

• Distance

Dynamic Range of Speech

The difference between the lowest level of speech i.e., speech minima and the highest level of speech i.e., speech maxima at any frequency

• Minima: 10%

• Maxima: 90%

• Dynamic Range: 10-90%

_______depends greatly on the length of a speech sample (Averaging longer samples=smaller dynamic range)

speech dynamic range

___ is the dynamic range suggested for making predictions about speech recognition

30 dB

Why is LTAAS useful?

can quantify the relationship between speech levels and hearing thresholds, giving us a specific indication of audibility for a given patient.

When combined with example of hearing loss, it tells us that speech below the plotted audiogram is audible and speech above the plotted audiogram is not audible

Articulation Index (AI):

Sound metric developed to indicate how much background sound levels can interfere with human speech, Tool used to predict the amount of speech that is audible to a patient with a specific hearing loss

AI has a value ranging from 0-100%

0%: no speech understood

100%: all speech understood

______ standards have been developed regarding AI measurement, including the current method, which is called the ______

ANSI, Speech Intelligibility Index (SII)

Articulation Index in Clinics

• Count the dot audiogram

• Many variations of this

• Count the dots below the threshold line

• The more dots that occur below the hearing threshold line, the more speech energy that the individual will perceive.

AI provides an indication of relative changes in audibility but is not useful for precise prediction of an individual patient’s speech recognition score.

Why?

• No 1:1 relationship between the AI and speech understanding

• 70% AI does not equal a 70% speech recognition score.