AudiologyOnline Phone: 800-753-2160

Interview with Wendy Davis M.S., AVR Sonovation

Wendy Davis, MS

July 26, 2000

AO/Beck: Hi Wendy, thanks for agreeing to tonight's interview. I'm looking forward to discussing your frequency transposition products as they represent an innovative and rather unique way to amplify sound. Nonetheless, I like to always start by asking a little about your education. What can you tell me?

Well, I received my master's degree in audiology from the University of North Carolina at Chapel Hill in 1989. Prior to obtaining my degree in audiology I was a speech language pathologist and I worked in southern-central Virginia as a speech
therapist in the public schools. Prior to that I was involved in business, primarily banking.

AO/Beck: When did you join AVR? And what does 'AVR' stand for?

AVR/Davis: I joined AVR Sonovation in 1994. AVR Sonovation is the 'daughter company' of AVR Communications, LTD, Israel. AVR Communications is responsible for research and product development of all AVR technology. AVR Sonovation, located in Eden Prairie, MN, is responsible for field trials of AVR technology, as well as Sales and Service of AVR products for North America.

AO/Beck: Stepping back for just a minute, what was the first product AVR introduced in the USA?

AVR/Davis: The first product was a vibrotactile unit, 'Trill', introduced in 1990. The second product was the 'TranSonic' Frequency Transposition Hearing Aid, introduced in 1991. TranSonic was the first wearable proportional frequency transposition hearing aid that operated in real-time. The TranSonic was a body hearing aid with an external microphone. Once the clinical benefit of AVR's proportional frequency transposition technology was documented, AVR started working on a behind-the-ear version to address the obvious limitations of a body aid. In 1998, AVR introduced ImpaCt DSR675 power BTE, the first Dynamic Speech Re-Coding (DSR) hearing aid. Dynamic Speech Re-Coding is based on two processing strategies, proportional frequency compression and dynamic consonant boost. With the introduction of ImpaCt, AVR decided to use the term frequency compression in place of frequency transposition because frequency compression is a more accurate description of what is occurring.

The basic concept behind AVR's Proportional Frequency Compression is the following: there is a digital phoneme analyzer that determines if the incoming speech segment is voiced or voiceless. For example, if the word 'sat' is spoken, the phoneme analyzer would determine that /s/ is voiceless, /a/ is voiced, and /t/ is voiceless. There are then three primary 'channels' the speech sound can travel through; the 'standard' hearing aid channel, the frequency compression for voiceless (FCVL) sound channel, and the frequency compression for voiced (FCV) sound channel.

The 'standard' hearing aid channel means that no frequency compression is applied to voiceless or voiced sounds. The input speech would move through the hearing aid like an input/output AGC system with filter shaping. The standard programmable hearing aid features include: AGC-I knee point, AGC-I release time, AGC-O, high pass filtering and low pass filtering.

The frequency compression for voiceless (FCVL) parameter is a programmable feature with a range of frequency compression from Disabled (1.0), which is no frequency compression to 5.0, which is the maximum FCVL. The smallest amount of FCVL available is FCVL of 1.5. The audiologist can program FCVL between 1.5 and 5.0 in 0.25 steps.

The frequency compression for voiced (FCV) parameter is also a programmable feature but it is either Disabled (1.0) or Enabled (1.25). This feature is typically used for only the most profound cases.

AO/Beck: So, if we're interested in the voiceless spectral energy at 4000 Hz, and we used FCVL of 4 - what would the effect be?

AVR/Davis: It's really simple mathematically; all you do is divide by the FCVL value. So, 4000 divided by 4 is 1000. If FCVL is set to 3.50, divide 4000 Hz by 3.5 and the new frequency where you would hear the 4000 Hz tone would be about 1142 Hz. Of course there are several factors to consider when determining the amount of FCVL. Slope of the hearing loss as well as actual thresholds are two parameters that assist ImpaCt's fitting software in determining initial FCVL values.

AO/Beck: What do these instruments sound like to those of us lucky enough to have normal hearing?

AVR/Davis: The first observation is that the voiceless speech sounds are lower in pitch. Some clinicians have also described the voiceless sounds as sounding 'lispy'. The important thing to remember though is the client will now be getting presence of speech cues where there used to be absence. The clients learn to incorporate this information and we typically see rapid improvement in speech production of adult users who have hearing-impaired sounding speech. For audiologists fitting the technology, the frequency-compressed sound takes some getting used to. It definitely sounds different than other hearing aids but the benefit to the hearing impaired person seems to make clinicians accept that 'different' is a good thing in this case.

AO/Beck: Wendy, can you give me an operational definition of 'frequency compression'?

AVR/Davis: Frequency Compression is the ability to proportionally compress critical speech information into lower frequencies in a selective manner. We compress in the frequency domain instead of just compressing in the amplitude domain. Critical high frequency speech information is 'moved' lower in frequency. For example, if we think about the spectral spread of the /sh/ sound - lets say the first energy band of /sh/ occurs in the 2000 Hz region, the second energy band in the 4500 Hz region, and the third energy band in the 6500 Hz region and so on. We decide to frequency compress voiceless sounds (FCVL) at a ratio of 2. The following happens to /sh/:
2000 Hz divided by 2= 1000 Hz
4500 Hz divided by 2=2250 Hz
6500 Hz divided by 2=3250 Hz
The entire speech band is frequency compressed. Because the entire band is compressed the /sh/ sound's identity is maintained because all energy bands have had the same amount (ratio) of frequency compression applied. We have proportionally lowered the sound.

Proportional, selective frequency compression can benefit many hearing impaired individuals because typically -- most patients have better hearing in the lower frequencies. By lowering critical speech sounds, we move into areas where patients have better residual hearing.

The need for an alternative to amplitude-based amplification strategies is also becoming more apparent with the work of Cynthia Hogan, Chris Turner, Teresa Ching, etc, indicating that when thresholds above 2000 Hz are 65 dB or poorer, additional amplification to these frequency regions does not improve speech recognition. This speaks to the issue of hair cell viability. So, frequency compression becomes an amplification option clinicians need to consider because critical speech information is still located in the high frequency regions. The ability to 'move' information so it can be used by the client is very exciting.

AO/Beck: Is 'frequency compression' the same as 'frequency transposition'?

AVR/Davis: In essence yes. The terms are used interchangeably. Presently the preferred term is frequency compression because it more accurately represents what is happening in AVR's current Dynamic Speech Re-Coding technology.

AO/Beck: Wendy, what does the term 'selective frequency compression' mean?

AVR/Davis: Selective frequency compression means that voiced sounds are processed independently from voiceless sounds. As mentioned earlier, in ImpaCt, there is a front-end analysis which does a voiced versus voiceless analysis. Then, based on how ImpaCt is programmed, one of the following selective processing strategies will take place:
  1. No frequency compression for voiced or voiceless sounds will be applied.

  2. Only frequency compression for voiceless (FCVL) sounds will be applied and voiced sounds will move through the hearing aid like any input/output AGC system with filer shaping.

  3. Both frequency compression for voiced sounds (FCVL) and frequency compression for voiceless sounds (FCV) will be applied.

  4. Only frequency compression for Voiced sounds will be applied and voiceless sounds will move through the hearing aid like any input/output AGC system with filter shaping.

There is another element that makes AVR's Dynamic Speech Re-Coding strategy selective. Two programmable features called Dynamic Consonant Boost (DCB) are also available. These controls provide additional gain to voiceless sounds. There is DCB1, which provides additional gain to higher frequency voiceless sounds, and DCB2, which provides additional gain to lower frequency voiceless sounds. DCB can be set independently of frequency compression or in combination with frequency compression. I short, we are analyzing the incoming speech segment and then determining what to do with it, based on that specific phoneme's characteristics.

AO/Beck: Does frequency compression occur in real-time?

AVR/Davis: With ImpaCt, the frequency compression occurs in about 2-4 milliseconds. The processing time can be a bit longer if there are high levels of environmental noise. With the TranSonic, ImpaCt's predecessor, the processing time was about 9 milliseconds. Perceptually this is still real time but some clients who used the TranSonic, especially good lip readers, commented that initially they sensed a slight delay. This 'feeling' went away after a few hours of using the TranSonic. We have had no reports of clients sensing a delay with ImpaCt.

AO/Beck: At first glance, it seems that if you were to 'load' the speech signal with low frequency sounds, you would get a tremendous 'upward spread of masking' and a high potential for speech to sound 'mumbled'? Is that correct?

AVR/Davis: The frequency compression is occurring is real time so we are not 'loading' or 'dumping' everything into the lows. When we talk, voiced and voiceless sounds do not occur simultaneously (unless you have multiple talkers.) If we go back to the example of the word 'sat', ImpaCt processes /s/, the client hears it; ImpaCt processes /a/, the client hears it; ImpaCt processes /t/, the client hears it. The hierarchy of speech (CVC, VCV patterns) is maintained. Another important point to mention is that it is not the absolute frequency of a sound which gives it its' identity. It is the relationship of the sound's energy bands (voiceless sounds) or formants (voiced sounds), which enables us to discriminate it from other sounds. Therefore, if we maintain the sound's energy band relationships by using proportional frequency compression, we maintain each sounds identity.

AO/Beck: With your system, can you apply frequency compression to just the higher pitches, or do you always compress the whole signal?

AVR/Davis: Good question, this causes a lot of confusion. Because we us a selective frequency compression process, the clinician can determine whether voiced sounds are compressed or voiceless sounds or compressed. Lets say that an audiologist decides to only compress voiceless sounds. If this is the case, then when a voiced sound enters ImpaCt, it will NOT be compressed but if a voiceless sound enters it will be compressed. The entire spectrum of the voiceless sound will be compressed.

AO/Beck: So the ImpaCt system will do an entire spectral frequency compression of the speech signal?

AVR/Davis: Yes, the whole speech signal is frequency compressed. Keep in mind that ImpaCt's digital phoneme analyzer is tracking each speech sound as it comes into the microphone of the hearing aid. so if /s/ enters, and frequency compression for voiceless has been programmed into the hearing aid, the entire spectrum compressed. This means everything from 0Hz to the end of the /s/ is compressed.

AO/Beck: As I learn more about the product and the technology I try to figure out who the ideal patient is. It appears to me that the target patient would be the ones with precipitous, severe to profound high frequency hearing loss?

AVR/Davis: Absolutely, they are the ones who are hardest (if not impossible) to reach with standard technologies. This population receives significant benefit from frequency compression technology. Imagine if patients with 90 and 100 dB losses in the high frequencies could actually perceive speech information located at 3000, 4000 and 6000 Hz.

A general rule of thumb, any person with thresholds of 65 dB or poorer above 2000 Hz is a potential candidate for frequency compression technology. The thresholds in the low frequencies can be in the normal to severe range.

AO/Beck: Additionally, because the input and output frequencies are of different spectral content, meaning the output is lower than the input, it seems like you can provide the high frequency information with considerably less feedback than we would traditionally expect when fitting severe to profound high frequency precipitous losses?

AVR/Davis: Absolutely. We have considerably less feedback problems when frequency compression is used because the input signal and output signal are at different frequencies. Another benefit of frequency compression technology is fewer problems for patients that have recruitment and tolerance issues. As we all know, recruitment and tolerance problems occur more with high frequency stimuli. Because we are moving information lower in frequency, we are able to move out of the areas where tolerance problems might occur. Because we typically 'move' information where a person has better residual hearing, we also need less gain and output than with traditional amplitude amplification strategies. This makes frequency compression technology not only beneficial for patients with tolerance problems but also very useful for pediatric populations where output is a concern. With frequency compression, we can provide access to critical high frequency speech information and use less gain and output while doing so.

AO/Beck: Can I ask about sales? How many of the frequency compression units have been sold?

AVR/Davis: The DSR675 high-power BTE that we've been selling for about two years has 8000+ units in the market worldwide.

AO/Beck: Is the technology available in an ITE?

AVR/Davis: Yes, the Dynamic Speech-Recoding features of frequency compression and dynamic consonant boost are available in a full shell ITE

AO/Beck: Are the all the technologies available as digitally programmable instruments?

AVR/Davis: Yes. All ImpaCt technologies are digitally programmable.

O/Beck: What about FM systems?

AVR/Davis: As most people are aware, AVR was the first company to introduce BTE/FM technology in 1991. Because we still believe that FM will always provide the best signal when a person is in a noisy environment, we have our own FM boot, FMB-20, that works with ImpaCt DSR675. FMB-20 attaches to the bottom of the DSR675 and enables the user to get the benefit both Dynamic Speech Re-Coding and FM. The operating frequencies are in the 216-217 MHz band.

We also recently introduced LogiCom-20 BTE/FM, which is an integrated BTE/FM that also has Dynamic Speech Re-Coding (frequency compression and dynamic consonant boost). The FM is inside the BTE case. No FM boot is required. This product will remove the concern that FM boots tend to fall off and get lost. With LogiCom-20, you get the ImpaCt circuit and FM capabilities all in one case.

AO/Beck: What about compatibility between different manufacturers of FM components?

AVR/Davis:Currrently, AVR and Phonak have FM boot technology and FM transmitters that operate on the 216-217 mHz band. Our products are compatible. You can use an AVR transmitter with Phonak receivers or a Phonak transmitter with FMB-20 boot for ImpaCt or with LogiCom-20 BTE/FM. The important thing is to let us know what frequency your existing technology is operating on. We can then make sure all equipment is frequency compatible.

DSR-675 also has an FM boot that operates on the 72-76mHz band. This makes it compatible with existing AVR BTE/FM, Extend-Ear. There is also an audio shoe that can be connected to DSR675 so it may be coupled to body-style FM via DAI.

AO/Beck: With the LogiCom 20 BTE/FM or when the FM boot is connected to ImpaCt DSR675, can you turn on the FM only, or the mic only as needed by the user?

AVR/Davis: Yes, there are three clinician programmable options on both products: hearing aid mic only, hearing aid mic plus FM, or FM only. With DSR675 you also have a programmable t-coil option for inductive couplings as well as an auxiliary option. With the auxiliary option, you would connect an audio shoe and can then interface with a variety of audio devices. All the hearing aids I have discussed provide a two, three or four program option. The clinician would determine the number of programs and listening options based on user need. The clients access the programs via a momentary switch. The switch can be configured to function differently for children or adults with dexterity problems.

AO/Beck: What can you tell me about the pricing of the FM system?

AVR/Davis: Our FM pricing is extremely competitive. I believe we are a little less 'pricey' than the other manufacturers and distributors of similar FM products. As you know, FM boot technology is considered expensive but this is due to the relatively small quantities of these products that are in the market. I anticipate cost will drop eventually as popularity increases.

AO/Beck: Without going into specifics, what can you tell me about the price of the technology?

AVR/Davis: The pricing falls in between high-end digitally programmable hearing aids and fully digital hearing aids. Most clinicians have been pleasantly surprised at the cost.

AO/Beck: Wendy, this is a fascinating product and it's refreshing to learn about a technology that approaches hearing health care in a unique way. I certainly wish you luck in this venture and I'd be happy to publish some clinical outcomes data from you in the near future!

AVR/Davis: Thank you Doug. I'll see if I can gather some information on outcomes and get that to Audiology Online over the summer so we can further the discussion regarding the patient's benefit.

AO/Beck: Thanks Wendy, I'll look forward to it.

Grand Rounds Live | 4 advanced live webinars | Register today

Wendy Davis, MS

Our site uses cookies to improve your experience. By using our site, you agree to our Privacy Policy.