"What did you say?" - A brief review of Speech Intelligibility
By: Zachery L'Italien, MCH Senior Acoustical Consultant
Have you ever had to ask, “What did you say?” Chances are you could hear the other person, but couldn't sufficiently understand what they were saying. This problem may have stemmed from conditions that interfered with good speech intelligibility, whether loud background noise, or an excessively reverberant environment.
The goal of good Speech Intelligibility is to clearly understand speech. A previous Tech Blog, STC – role in Speech Privacy (mchinc.com), by Tony Hoover reviewed several considerations for Speech Privacy design, which is largely the inverse (i.e., inhibiting the ability to understand speech).
Speech Intelligibility, like Speech Privacy, depends on Speech Level and Background Noise Levels. Speech Intelligibility is typically considered for applications where the “Source” (the talker) and the “Receiver” (the listener) are in the same space. Therefore, Sound Isolation is not discussed here, although Reverberation is.
The three most important factors for Speech Intelligibility are:
Speech Level of the “Source” (the talker)
We generally categorize speech levels as “normal”, “raised”, or “loud”. Loud speech (roughly 72 dBA) usually offers greater Speech Intelligibility than normal (roughly 60 dBA). Of course, we tend to raise our voices if, at first, we aren't understood. Occasionally, speech is amplified through an electronic sound system, which can enhance Speech Intelligibility if properly designed.
Background Noise Levels
Background noise levels in the room, from the HVAC system or miscellaneous noises, "mask" speech sound. Quiet background noise levels are beneficial to Speech Intelligibility.
Reverberation is how long a sound persists in a space after it is suddenly stopped. In general, shorter reverberation times (T60) are desirable for good Speech Intelligibility. T60 is influenced by both the room volume and absorptive finishes.
In order to achieve good Speech Intelligibility, all three factors must be considered as a system. Good Speech Intelligibility is key in public address spaces, conference rooms and lecture halls, among many others.
Consider an instructor and students in an unamplified lecture hall where reverberation time is adequately minimized.
In this case, Speech Intelligibility between the instructor and students is largely dependent on Speech Level and Background Noise.
If the background noise is quiet enough for lectures, and an instructor lectures in a raised "presenter" voice (say 35dBA above background), they should be readily understood by students.
This results in "excellent" intelligibility.
If a student asks a question in a normal voice, they should still be clearly understood, albeit not as well as the instructor.
This results in "fair" intelligibility.
Now let’s consider the same scenario, but in this example the HVAC noise was not appropriately controlled, and is generally too loud. In this case, the instructor’s loud speech is still understood. However, the student’s normal speech is unintelligible, which in many cases in unacceptable.
There can be significant negative implications if a space, where communication is highly important, is not appropriately designed for good Speech Intelligibility. Talk to your acoustical consultant about how to achieve appropriate speech intelligibility in a cost-effective manner.