Directory  |  Calendar  |  Forms  |  Contact Us  |  Login

The Hearing Professional Magazine

Search Keywords:

Search By:

Article Submission

This article originally appeared in The Hearing Professional November—December 2004

Related Links

The I/O Curve—Friend or Foe?

By Mary Leisses, MS

IHS offers a diversity of options for obtaining continuing education credit: seminars and classroom training, institutional courses, online studies and distance learning programs. This article represents yet another opportunity. Upon successful completion of the accompanying test you will earn one CEU.

(Please check with your licensing board to verify acceptance of continuing education credit for this activity.)

Digital technology has added a new dimension of flexibility in fitting and meeting the needs of our patients. As digital moves into increasingly complicated algorithms and combinations of parameters, we rely more and more on the manufacturer-provided “quick fit” and “fitting assistants.” As we proceed on this path, it’s important that we do not lose sight of how the various parameters work and interact. Even in the age of advanced digital, it is still imperative to understand the basics of the input/output (I/O) curve. In this article, we will break down the I/O curve and discuss the rationale, purpose and guidelines for each of the components.

Background Vocabulary

The following definitions will ensure we are all using the same basic terms correctly.

Input/output function is a graph or curve illustrating how the output of a circuit changes relative to the input. It is typically done one frequency at a time, though a composite signal may also be used. The range of input will vary from 50 to 90 dB if the test is done according to ANSI guidelines.

Input compression refers to a compression scheme that considers the input level to determine if the compression should be activated. The check for compression will come before the volume control (feed forward loop). This means that adjustment of the volume control will affect both gain and output.

Figure 1. The top diagram illustrates a basic analog circuit.

Output compression refers to a compression scheme that considers the output level (input level + gain) to determine if the compression should be activated. The check for compression will come after the volume control (feedback loop). This means that adjustment of the volume control will affect only gain. The maximum output of the circuit is determined by the compression setting.

Compression ratio characterizes the amount of compression or automatic gain adjustment that will occur. The formula for calculating ratio is change in input / change in output. If your input level increases 20 dB while your output level increases 10 dB (20/10 = 2) your compression ratio is 2:1. Ratios are always expressed relative to 1. For every 2 dB increase in input, you will have a 1 dB increase in output. Another example shows a 20 dB increase in input with only a 2 dB increase in output. This would be a 10:1 compression ratio (20/2 = 10).

Figure 2. Two different compression ratios with their formula. The diagram illustrates the feed forward loop used in input compression circuits.

Linear compression could have two different meanings: 1) a 1:1 compression ratio, meaning linear gain and 2) a compression scheme where the compression ratio is fixed. For example, it will always be 1.8:1 as long as you are in compression. The opposite of this would be curvilinear compression.

Figure 3. The difference between linear compression (a fixed compression ratio) and curvilinear compression. The diagram illustrates the feedback loop used in output compression circuits.

Curvilinear compression is a type of compression where the ratio varies with the input level. Typically, as the input level increases, the compression ratio also increases.

Compression threshold (kneepoint) is the point at which the circuit will go into compression. Threshold and kneepoint are often used interchangeably.

Attack time is the length of time it takes for the threshold to be exceeded and the circuit knows to go into compression.

Release time is the amount of time the circuit stays in compression and until it is within 2 dB of its linear gain once the signal is gone.

Time constants are the attack and release times together. Time constants may be fast, slow or both fast and slow, depending upon the circuit and signal.

Figure 4. The compression threshold is commonly called the kneepoint because it looks like a bent knee on the I/O graph. This graph shows a low and high threshold.

Syllabic compression uses fast attack and release times. The goal of this configuration is to compress the vocalized elements of speech (vowels) while allowing the softer consonant sounds to pass through with linear amplification. The theoretical result would be an increase in the consonant-vowel ratio.

Dual-time constants allow the circuit to have more than one-time constant configuration. For example, very loud inputs would utilize a fast attack time, while softer inputs would use a slower attack time. The release times can also be varied by the algorithm.

Detection system defines how the system knows the threshold has been exceeded for activation of the compression. The very first compression systems used “peak” detectors that required only one signal to exceed the threshold to activate the compression. Later, circuits used various averaging techniques. These allowed the inputs over a specified time period to be averaged and then compared to the threshold. This provided a more realistic representation of the environment and prevented short bursts from shutting the systems down.

Wide-Dynamic-Range Compression

Wide-dynamic-range compression (WDRC) is a hearing aid compression algorithm with a low threshold of activation, designed to deliver signals between a listener’s thresholds of sensitivity and discomfort in a manner that matches loudness growth. Basically, we want all input levels (soft to loud) to be heard by the hearing impaired individual in the same relative way a normal hearing person would. Therefore, soft speech, around 35 dB SPL, would be perceived as very soft to the hearing impaired individual and not necessarily easy to understand. Conversational speech should be comfortable and easy to understand. Loud sounds and environments should be perceived as loud but not uncomfortable.

To achieve this goal WDRC will have a low compression threshold; in some digital products it is as low as 20 dB SPL. The highest a WDRC threshold will typically go is 55 dB SPL. The low threshold will allow a good amount of linear gain to be applied to very soft inputs so that they can be brought into the auditory range of the individual.

Let’s say our circuit has the threshold at 35 dB SPL. All sounds below 35 dB SPL would receive linear gain. From 35 dB up, all the inputs would be under compression. Because such a wide range of inputs is under compression, the ratio is usually pretty small. The highest WDRC ratio is usually not more than 4:1. Think of it like an accordion. In order to fit the accordion into the box, we are compressing each section of the middle just a little. In order to fit the full range of sounds into the narrowed dynamic range of the hearing impaired individual, we are applying a small amount of compression to a large range of inputs.

When looking at different products, what do you want in a WDRC type circuit? Ideally, the ability to adjust thresholds and ratios independently of each other will provide good flexibility. In some products, you may not have direct access to ratio. In these products, by adjusting the gain for soft or quiet sounds and gain for loud sounds, you will affect the ratio.

Some products allow you to select different time constant options; however, there is not an abundance of research on when to use what time constants. Some research points toward slower time constants being preferred by individuals, particularly relative to speech level inputs. Other research has shown that fast-time constants (syllabic compression) in the low frequencies coupled with slower constants in the high frequencies can also be acceptable for individuals.


WDRC does an excellent job of amplifying soft sounds. It does so well, in fact, that often users of hearing aids are annoyed by the soft environmental sounds that surround us on a daily basis. The purpose of expansion is to bury or hide these environmental sounds below the threshold of the individual.

Expansion is greater than linear gain. Therefore, if linear is 1:1 (for every 1 dB increase in input there is 1 dB increase in output) and compression is 2:1 (for every 2 dB increase in input there is a 1 dB increase in output) expansion is .5:1 (for every half dB increase in input there is full dB increase in output).

Figure 5. Linear, compression and expansion are illustrated on the same I/O graph.

So why do we need greater than linear gain? When expansion is activated in the hearing aid, the hearing aid will ignore sounds below a specified input level, at about 20 dB. That means that sounds softer than 20 dB are neither amplified nor compressed. They merely pass through the system. Once a signal level is 20 dB or above, amplification will be provided.

Now the problem. If our threshold for compression is at 50 dB, that is our target gain for soft speech. If we use linear gain from 20 dB to up to our threshold we will not have enough gain to make speech audible. Therefore, we need greater than linear gain in order to “ramp up” quickly. This allows the circuit to bypass very soft sounds, provide less than linear gain to many intermediate sounds and still provide enough gain for soft speech to be audible.

Figure 6. If linear gain was used when “hiding” sounds below the functioning level of the hearing aid, the target gain for soft speech would never be used. Expansion allows minimal gain for soft environmental sounds while still allowing the need gain for speech.

What features do you want in your products? You want the ability to turn off the expansion because not everyone will like it. Previous long-time users of hearing instruments without expansion or those with more moderate losses might perceive that they are out of touch when expansion is on. The hearing instrument should have the capability for expansion to be on in one memory and off in another. For example, in a music memory you might want expansion turned off so that the listener does not miss the softer portions of musical pieces. The same individual might want expansion to be on in the primary listening memory.

Compression Limiting

With the goal of limiting distortion and controlling output, compression limiting was the first type of compression used on a regular basis in hearing aids. Compression limiting can be either input or output compression. Most often, it is configured as output compression and the two terms are often used interchangeably. Compression limiting is characterized by a high threshold and a high ratio.

In today’s products, compression limiting is usually combined with WDRC. While WDRC is typically dealt with at the channel level, the compression limiting configuration varies from product to product. It is fairly common to have the limiting be broadband, meaning across the entire frequency range instead of channel specific. With compression limiting you need the flexibility to adjust the maximum output of the instrument by 10 to 15 dB. Most digital products cannot be made absolutely linear, so that is an unreasonable expectation.

In Conclusion

As digital moves forward it is important to understand the basics of the algorithms utilized. Understanding the I/O curve is the first step in providing successful fittings for your patients. A working knowledge of this basic level of the algorithm provides the foundation and framework for the more advanced features of digital technology. THP

Mary Leisses, MS, is a staff audiologist at St. Croix Medical in Minneapolis, Minnesota. Please direct communications to her at (Ms. Leisses thanks Lois Benson for her assistance with this article.)


ANSI S3.22-1987 American National Standard Specification of Hearing Aid Characteristics. New York: Acoustical Soc of Am, April, 1987.

ANSI S3.22-1996 American National Standard Specification of Hearing Aid Characteristics. New York: Acoustical Soc of Am, January, 1996.

Byrne D, Dillon H: The national acoustic laboratories (NAL) new procedure for selecting the gain and frequency response of a hearing aid. Ear and Hearing 1986; Vol 7, No 4.

Byrne D: Hearing aid selection for the 1990’s: where to? J Am Acad Audio, Vol 7, No 6.

Delk JH: Comprehensive Dictionary of Audiology. MA: The Laux Company 1983.

Dillon H: Tutorial: Compression? Ear and Hearing 1996, Vol 17, No 4.

Fortune TW, Preves D A, Woodruff BD: Saturation-induced distortion and its effects on aided LDL, Hearing Instruments 1991, Vol 42, No 10, pp 37.

Hawkins D: Selection of hearing aid characteristics, Hearing Aid Assessment and Use in Audiologic Habilitation, 3rd ed 1986, Ed WR Hodgson, Baltimore: Willliams and Wilkins, 128—1.

Killion MC: A critique on four popular statements about compression, The Hearing Review 1997, Vol 4, No 2, pp 36 and 38.

Killion, MC: Compression: Distinctions, The Hearing Review 1996, Vol 3, No 8, pp 29—32.

Kuk F: Theoretical and practical considerations in compression hearing aids, Trends in Amplification 1996, Vol 1, No 1.

Libby ER: Hearing instrument systems technology and performance standards, Hearing Instrument Science and Fitting Practices 1st ed 1985, Ed RE Sandlin, MI: IHS 201—300.

Skinner MW: Hearing Aid Evaluation 1988, New Jersey: Prentice Hall, 220—3.