Interview with Greg Ernst, President, Sonovation Inc.
Share:
Topic: A Primer on Digital Frequency Compression
Dybala: Greg, thanks for being here today. I think we have some interesting things to talk about regarding Sonovation and Digital Frequency Compression in hearing aids.
Ernst: Thank you for having me, glad to be here.
Dybala: Just to kick things off, let's first talk about you. Could you give us a little bit of your professional background?
Ernst: Glad to, Paul. I entered the industry when Tony Lang, president of Lang Hearing Instruments invited me to join him in the early '80s. I had studied electronics and worked on radio controlled off-peak electrical systems. I put that knowledge to work in circuit design and simultaneously began learning about hearing aid fitting. I worked for a time at Argosy on their design team. In 1990, I became president of Active Electronics and subsequently met and began working with Barak Dar of Sonovation. Barak was the creative spark behind many of Sonovation's ground-breaking products, including Frequency Compression. In 2001, I joined Sonovation and in 2005 I assumed the position of president.
Dybala: Sonovation is known for its innovative hearing aid technology, Digital Frequency Compression (DFC). Give us a quick description of what this hearing aid technology is and what it actually does.
Ernst: Digital Frequency Compression is an algorithm that runs on a 17 band, 9 channel open platform DSP. The DFC algorithm analyzes the incoming speech signal and determines if the speech is a voiced sound or a voiceless sound. If the sound is voiced, it is amplified based on the gain/output parameters required for the low-to-mid frequency hearing loss. If the sound is voiceless, the DFC algorithm can "move" the voiceless sounds to lower frequency regions where most clients have better - and more useable residual hearing.
Dybala: This is a pretty unique processing strategy in the industry. Are there any other manufacturers who provide this type of technology?
Ernst: To date, Sonovation is the only company with this processing strategy and we are about 20 years ahead of the game, but we understand Widex will be introducing a hearing instrument with a frequency transposition algorithm at AAA.
Dybala: Just for my own semantic gymnastics, when you say "frequency compression" you are talking about "frequency transposition", correct? I just mention this as when I heard the word compression I think of changes in output level in hearing aids.
Ernst: You are correct that many people use the terms interchangeably. Sonovation prefers the term frequency compression because it more accurately represents what Sonovation has historically done with the speech signal.
Dybala: Over the past few years one of the "buzz words" in audiology has been Cochlear Dead Regions, a term used by Dr. Brian Moore and colleagues. I know many of us have also worked with patients who have a hearing loss that changes from a mild or moderate loss in the low-to-mid frequencies to the severe or even profound level in the high frequencies. Are these patients good candidates for DFC?
Ernst: Absolutely. Many of these individuals have high-frequency hearing losses that tradition amplification cannot effectively deal with. Either the high-frequency thresholds are so poor that traditional amplification just can't deliver sufficient gain/output to make spectral energy cues of voiceless sounds audible, or the "dead" regions exist and the client can't use the high-frequency signal even if the hearing could deliver it. When the high-frequency signal cannot be delivered, speech intelligibility cannot improve.
There is also a myth in the hearing industry that aggressive noise reduction algorithms, take care of the speech intelligibility issues. This is just not correct for a large segment of the hearing impaired population. If the client can't detect and more importantly can't recognize the differences between voiceless speech sounds like /s/, /t/, /sh/, /k/, /f/, etc, in a quiet environment with their current amplification - there is no noise algorithm that is going to compensate for the inability to hearing these sounds.
Another critical issue is that audiologists/dispensers will determine the amplification goals based on the slope of the hearing loss, acknowledging that they cannot amplify the high-frequency speech information due to the severity of the loss. This is also an acknowledgement that speech understanding will be compromised. At Sonovation we understand that patients don't need to compromise. Our slogan, "Move it, don't lose it!" means voiceless speech sounds and improved intelligibility are available. The clinician just needs to think outside of the box.
Dybala: Couldn't someone argue that by moving those sounds from a higher to a lower frequency region could cause more problems that it would solve?
Ernst: The answer to this question is no. Two important points:
Dybala: Can you expand on those points for our readers?
Ernst: Sure! Just to paraphrase your original question, people will commonly ask, "Does DFC create noises for some patients, hindering their speech recognition?" I think this concept of "creating sound" is where DFC gets misunderstood. Frequency compression does not create sounds - it compresses the needed voiceless consonants to more useable low frequencies, to increase their audibility and recognition. The patient is hearing existing sounds in a new way. This brings up an important counseling issue. If a sound has never been heard, or has not been heard in a long time, the brain will automatically classify this sound as "noise" until meaning has been attached. Here is one way we suggest that an audiologist/dispenser might counsel a patent:
Ernst: The compressed sounds have the same primary structure as the original sounds, but since different hair cells are being stimulated, they may have an unfamiliar sound quality compared to the originals. The patient must learn to use the newly audible, compressed voiceless sounds by attaching meaning, i.e., context. The newly audible sounds command attention, but may not be immediately understood, thus the patient may initially regard them as interfering with understanding. The compressed voiceless sounds will ultimately improve understanding while reducing effort, but the patient must help the acclimation process along with patience and auditory rehabilitation.
Dybala: Do you have any audio examples of what speech sounds like when processed by DFC?
Ernst: Yes, I do. Here are two samples of the sentence, "The basket is full of grapes". While the second speech sample has been processed with DFC, the sentence is still highly intelligible
Greg, you also mentioned auditory rehabilitation. What type of auditory rehabilitation would Sonovation recommend for new users of DFC?
Ernst: It is really a simple process. We recommend that you have your patients read aloud three or four times a day for three minutes each time. When a patient can connect the visual cue with the frequency-compressed sounds in their own voice, they are on their way to recognizing those sounds in other people's voices. Reading aloud expedites this process. If you would like additional suggestions, see the ImpaCt Workbook on our web site at
avrsono.com
Dybala: What are the types of specific audiometric criteria that we would want to consider in a patient for fitting a person with DFC?
Ernst: A patient may meet any or all of the following criteria:
Ernst: Exactly. There are certain comments that you hear from patients with traditional amplification and professionals know exactly what to do to help. For example, "I sound like I am in barrel" or "People's voices sound harsh or tinny".
One of the biggest challenges we face at Sonovation is that the type of comments that patients make when they are first adjusting to DFC are not what most professionals are used to dealing with! We work very hard to provide professionals with the training they need to make the right adjustments.
Dybala: What are some examples of this?
Ernst: A typical comment that a new user may make is that they hear a "trailing static" sound at the end of words. Patients with a severe high-frequency hearing loss often experience a gap in words where the inaudible (to them) high-frequency sounds occur. While Frequency Compression makes those missing sounds audible again, without context, the newly audible sounds may be misidentified as "static".
Dybala: So, this is really more of a counseling issue than a need to make adjustments to the hearing aid, correct?
Ernst: It could be both. This is a counseling issue as there is a need for the clinician to correctly identify the newly audible sounds for their patients as sought-after voiceless consonants. But, it is also good to check the gain/output parameters as this complaint could arise from excessive gain or output in the low frequency region where the voiceless sounds have been compressed.
Dybala: Interesting, my first inclination would have been to adjust the high-frequency regions and not the low.
Ernst: That is a common misunderstanding. You have to remember that the high-frequency voiceless sounds have been shifted to a lower frequency due to DFC.
Dybala: Ok, now that makes sense. I have to perform a little frequency compression on my fitting logic to help understand how DFC works!
Ernst: Exactly! It is really not complicated, just different. Once you get in the correct mindset, the rationale is quite intuitive. We find that with many of our first time customers we need to help them with the first couple of fittings, but after that the professional is up and running.
Dybala: Yes, it does make more sense to me now! Well, we are getting to the end of our session and so before we close, I wanted to mention your company web site at www.avrsono.com - as it has additional information on DFC and I would encourage the reader to take a look if they are interested in this technology.
I would also encourage our readers to visit the Audiology Online Ask the Expert section where we will provide answers to some common fitting questions on Sonovation DFC submitted by our members, we also have an interview that goes into the more technical aspects of your technology (Davis, 2000) in addition to the aforementioned Sonovation e-Learning courses in the AO e-Learning section (Davis, 2005). Thank you again Greg, for sharing this information with our readers today.
Ernst: Glad to do it!
--------------------------
Sonovation's dedication to improving the lives of individuals with hearing loss by introducing cutting edge technology has resulted in a history of hearing industry "firsts." Sonovation's technological breakthroughs, revolutionary product introductions and commitment to innovation have placed us in the forefront of hearing health care. Visit www.avrsono.com for more information.
Works Cited:
Davis, W. (2000, July 26) Interview with Wendy Davis. Audiology Online, Interview 9. Retrieved February 25, 2006, from the Interview Archives on www.audiologyonline.com.
Davis, W. (2005, September 1) Digital Frequency Compression (Transposition) made easy. Audiology Online, Recorded Course 4146. Retrieved February 25, 2006, from the e-Learning section on www.audiologyonline.com.
Ernst: Thank you for having me, glad to be here.
Dybala: Just to kick things off, let's first talk about you. Could you give us a little bit of your professional background?
Ernst: Glad to, Paul. I entered the industry when Tony Lang, president of Lang Hearing Instruments invited me to join him in the early '80s. I had studied electronics and worked on radio controlled off-peak electrical systems. I put that knowledge to work in circuit design and simultaneously began learning about hearing aid fitting. I worked for a time at Argosy on their design team. In 1990, I became president of Active Electronics and subsequently met and began working with Barak Dar of Sonovation. Barak was the creative spark behind many of Sonovation's ground-breaking products, including Frequency Compression. In 2001, I joined Sonovation and in 2005 I assumed the position of president.
Dybala: Sonovation is known for its innovative hearing aid technology, Digital Frequency Compression (DFC). Give us a quick description of what this hearing aid technology is and what it actually does.
Ernst: Digital Frequency Compression is an algorithm that runs on a 17 band, 9 channel open platform DSP. The DFC algorithm analyzes the incoming speech signal and determines if the speech is a voiced sound or a voiceless sound. If the sound is voiced, it is amplified based on the gain/output parameters required for the low-to-mid frequency hearing loss. If the sound is voiceless, the DFC algorithm can "move" the voiceless sounds to lower frequency regions where most clients have better - and more useable residual hearing.
Dybala: This is a pretty unique processing strategy in the industry. Are there any other manufacturers who provide this type of technology?
Ernst: To date, Sonovation is the only company with this processing strategy and we are about 20 years ahead of the game, but we understand Widex will be introducing a hearing instrument with a frequency transposition algorithm at AAA.
Dybala: Just for my own semantic gymnastics, when you say "frequency compression" you are talking about "frequency transposition", correct? I just mention this as when I heard the word compression I think of changes in output level in hearing aids.
Ernst: You are correct that many people use the terms interchangeably. Sonovation prefers the term frequency compression because it more accurately represents what Sonovation has historically done with the speech signal.
Dybala: Over the past few years one of the "buzz words" in audiology has been Cochlear Dead Regions, a term used by Dr. Brian Moore and colleagues. I know many of us have also worked with patients who have a hearing loss that changes from a mild or moderate loss in the low-to-mid frequencies to the severe or even profound level in the high frequencies. Are these patients good candidates for DFC?
Ernst: Absolutely. Many of these individuals have high-frequency hearing losses that tradition amplification cannot effectively deal with. Either the high-frequency thresholds are so poor that traditional amplification just can't deliver sufficient gain/output to make spectral energy cues of voiceless sounds audible, or the "dead" regions exist and the client can't use the high-frequency signal even if the hearing could deliver it. When the high-frequency signal cannot be delivered, speech intelligibility cannot improve.
There is also a myth in the hearing industry that aggressive noise reduction algorithms, take care of the speech intelligibility issues. This is just not correct for a large segment of the hearing impaired population. If the client can't detect and more importantly can't recognize the differences between voiceless speech sounds like /s/, /t/, /sh/, /k/, /f/, etc, in a quiet environment with their current amplification - there is no noise algorithm that is going to compensate for the inability to hearing these sounds.
Another critical issue is that audiologists/dispensers will determine the amplification goals based on the slope of the hearing loss, acknowledging that they cannot amplify the high-frequency speech information due to the severity of the loss. This is also an acknowledgement that speech understanding will be compromised. At Sonovation we understand that patients don't need to compromise. Our slogan, "Move it, don't lose it!" means voiceless speech sounds and improved intelligibility are available. The clinician just needs to think outside of the box.
Dybala: Couldn't someone argue that by moving those sounds from a higher to a lower frequency region could cause more problems that it would solve?
Ernst: The answer to this question is no. Two important points:
- Sonovation uses a selective proportional process to lower the frequency of the voiceless sounds.
What I mean by this is, only voiceless sounds are compressed. Voiced sounds are not compressed. As long as the voiced sounds are unaltered, the vowel transitions are not affected in anyway. This allows the hearing impaired person to receive the voiced components of speech as they do with any hearing aid. We are just compressing the voiceless sounds into a region where they can hear them. - As mentioned above, the frequency compression process Sonovation uses is proportional. When a voiceless sound has been identified by the phoneme analyzer in the hearing aid, the entire spectrum, 0-10 kHz, is compressed. This means that the energy peak relationships within a sound are maintained. It is the energy relationships that make speech sounds unique, not the absolute frequency where the sounds occur. This is a critical element in understanding how Sonovation's Frequency Compression technology works.
Dybala: Can you expand on those points for our readers?
Ernst: Sure! Just to paraphrase your original question, people will commonly ask, "Does DFC create noises for some patients, hindering their speech recognition?" I think this concept of "creating sound" is where DFC gets misunderstood. Frequency compression does not create sounds - it compresses the needed voiceless consonants to more useable low frequencies, to increase their audibility and recognition. The patient is hearing existing sounds in a new way. This brings up an important counseling issue. If a sound has never been heard, or has not been heard in a long time, the brain will automatically classify this sound as "noise" until meaning has been attached. Here is one way we suggest that an audiologist/dispenser might counsel a patent:
"I can't get you to hear the most critical sounds in speech in a tradition manner. If you want to improve your speech understanding, I have to deliver the important sounds to a new area in your cochlea. So, you now have a "new" /s/, /f/, /t/, etc."Dybala: So, what do frequency compressed speech signals sound like?
Ernst: The compressed sounds have the same primary structure as the original sounds, but since different hair cells are being stimulated, they may have an unfamiliar sound quality compared to the originals. The patient must learn to use the newly audible, compressed voiceless sounds by attaching meaning, i.e., context. The newly audible sounds command attention, but may not be immediately understood, thus the patient may initially regard them as interfering with understanding. The compressed voiceless sounds will ultimately improve understanding while reducing effort, but the patient must help the acclimation process along with patience and auditory rehabilitation.
Dybala: Do you have any audio examples of what speech sounds like when processed by DFC?
Ernst: Yes, I do. Here are two samples of the sentence, "The basket is full of grapes". While the second speech sample has been processed with DFC, the sentence is still highly intelligible
- The basket is full of grapes - unprocessed
Greg, you also mentioned auditory rehabilitation. What type of auditory rehabilitation would Sonovation recommend for new users of DFC?
Ernst: It is really a simple process. We recommend that you have your patients read aloud three or four times a day for three minutes each time. When a patient can connect the visual cue with the frequency-compressed sounds in their own voice, they are on their way to recognizing those sounds in other people's voices. Reading aloud expedites this process. If you would like additional suggestions, see the ImpaCt Workbook on our web site at
avrsono.com
Dybala: What are the types of specific audiometric criteria that we would want to consider in a patient for fitting a person with DFC?
Ernst: A patient may meet any or all of the following criteria:
- Thresholds 65-70 dB or poorer at 2KHz and above.
- Greater than 35 dB per octave slope.
- Inability to detect and/or recognize /s/, /sh/.
- Reduced speech recognition scores.
- Poor tonal perception.
- Hearing loss due to ototoxicity.
Ernst: Exactly. There are certain comments that you hear from patients with traditional amplification and professionals know exactly what to do to help. For example, "I sound like I am in barrel" or "People's voices sound harsh or tinny".
One of the biggest challenges we face at Sonovation is that the type of comments that patients make when they are first adjusting to DFC are not what most professionals are used to dealing with! We work very hard to provide professionals with the training they need to make the right adjustments.
Dybala: What are some examples of this?
Ernst: A typical comment that a new user may make is that they hear a "trailing static" sound at the end of words. Patients with a severe high-frequency hearing loss often experience a gap in words where the inaudible (to them) high-frequency sounds occur. While Frequency Compression makes those missing sounds audible again, without context, the newly audible sounds may be misidentified as "static".
Dybala: So, this is really more of a counseling issue than a need to make adjustments to the hearing aid, correct?
Ernst: It could be both. This is a counseling issue as there is a need for the clinician to correctly identify the newly audible sounds for their patients as sought-after voiceless consonants. But, it is also good to check the gain/output parameters as this complaint could arise from excessive gain or output in the low frequency region where the voiceless sounds have been compressed.
Dybala: Interesting, my first inclination would have been to adjust the high-frequency regions and not the low.
Ernst: That is a common misunderstanding. You have to remember that the high-frequency voiceless sounds have been shifted to a lower frequency due to DFC.
Dybala: Ok, now that makes sense. I have to perform a little frequency compression on my fitting logic to help understand how DFC works!
Ernst: Exactly! It is really not complicated, just different. Once you get in the correct mindset, the rationale is quite intuitive. We find that with many of our first time customers we need to help them with the first couple of fittings, but after that the professional is up and running.
Dybala: Yes, it does make more sense to me now! Well, we are getting to the end of our session and so before we close, I wanted to mention your company web site at www.avrsono.com - as it has additional information on DFC and I would encourage the reader to take a look if they are interested in this technology.
I would also encourage our readers to visit the Audiology Online Ask the Expert section where we will provide answers to some common fitting questions on Sonovation DFC submitted by our members, we also have an interview that goes into the more technical aspects of your technology (Davis, 2000) in addition to the aforementioned Sonovation e-Learning courses in the AO e-Learning section (Davis, 2005). Thank you again Greg, for sharing this information with our readers today.
Ernst: Glad to do it!
--------------------------
Sonovation's dedication to improving the lives of individuals with hearing loss by introducing cutting edge technology has resulted in a history of hearing industry "firsts." Sonovation's technological breakthroughs, revolutionary product introductions and commitment to innovation have placed us in the forefront of hearing health care. Visit www.avrsono.com for more information.
Works Cited:
Davis, W. (2000, July 26) Interview with Wendy Davis. Audiology Online, Interview 9. Retrieved February 25, 2006, from the Interview Archives on www.audiologyonline.com.
Davis, W. (2005, September 1) Digital Frequency Compression (Transposition) made easy. Audiology Online, Recorded Course 4146. Retrieved February 25, 2006, from the e-Learning section on www.audiologyonline.com.
