Interview with Wendy E. Davis Audiologist & Director of Professional Services, Sonovation Inc.
Share:
Topic: Frequency Compression (a.k.a. "frequency transposition")
Beck: Good Morning Wendy. It's great to speak with you again.
Davis: Hi Doug, nice to be with you too.
Beck: If you don't mind, I'd like to start by having you explain about Sonovation's unique hearing aid product. It's actually very different from typical analog and digital hearing aids, and I wonder if you can give us an overview of the product itself, and also of Sonovation?
Davis: Sure Doug. Sonovation is a subsidiary of AVR Communications. Many people are used to looking for us under the name AVR Sonovation, but last year we started using our official U.S. name, Sonovation. Sonovation/AVR has always been associated with cutting edge technologies. Our approach has been to examine the needs of hearing aid users, assess what was currently available in the marketplace and develop products that address the unmet needs of the deaf and hard of hearing. Development has always been from the ground up, no restraints and no assumptions.
Beck: That's very exciting...and if I recall, your initial target market was the severe-to-profound hearing-impaired population - right?
Davis: Yes, that's where Sonovation/AVR started. We knew that audibility and ultimately recognition of voiceless sounds for people with severe-to-profound hearing loss was a real challenge. Even though hearing aids made environmental sounds and vowel (voiced) sounds louder, hearing aids generally didn't have a way to present the softest, highest frequency speech sounds, without saturating the hearing aid circuit, causing distortion, or creating acoustic feedback.
Beck: Sure, and that used to be a very common and enormously frustrating issue. By the time the high pitched sounds were loud enough to be perceived, they were too loud for the hearing aid, and the proximity of the microphone to the receiver, with adequate venting and enough high frequency amplification, and pretty soon we had feedback, or we turned down the amplification in the highs....which was a lot like a dog chasing it's tail....we just couldn't get there! Of course, 90 percent of the intelligibility of English is based on the high pitched sounds, so if you cannot hear those sounds, people sound like they're mumbling!
Davis: That's right, this was and continues to be a common dilemma, Sonovation/AVR had to "think outside the box." We knew the high frequencies were critical for improving speech understanding. We knew that voiceless sounds were the most difficult sounds to amplify through hearing aids without creating additional problems. We also knew that hearing is typically poorer in the high frequencies. To effectively address these issues, Sonovation/AVR
developed an algorithm to proportionally move high-frequency "voiceless" sounds
to lower-frequency regions. This is where patients typically have better hearing and this is where the hearing aid is more efficient at amplifying. This approach has been called "frequency transposition" but we now use the term "digital frequency compression."
Beck: Very interesting. And so sounds like /m/ and /ee/ and other voiced sounds are not affected while the voiceless sounds /sh/ and /s/ are compressed into the lower frequency range?
Davis: Exactly. Voiced sounds are amplified like all other hearing aids, while the voiceless sounds are moved to lower frequencies, where the patient can hear them. This process is done is real time, so there is virtually no delay. We're also following the dynamics of speech; vowel, consonant, vowel (VCV); consonant, vowel, consonant (CVC). Voiceless sounds are not "dumped" on top of voiced sounds. The process we are using is proportional so we are maintaining the spectral relationships within and between voiceless sounds. These relationships are what lets us identify that an "s" is different from an "sh" is different from a "t" and so on.
Beck: So if I had a sound at 3000 Hz, and I wanted to compress it down to 1500 or 750 Hz, would I just put in a mathematical number, like "compress this into half it's current value" or a quarter of it's value or whatever?
Davis: Sort of. We use the term "digital frequency compression" (DFC) ratio. This ratio is really a divisor from 1.25 to 5.0 in 0.25 steps.
Beck: So you could take a sound of 6000 Hz, and make that sound compress into 750 Hz without too much difficulty?
Davis: No, not really. The maximum we frequency compress is by a ratio (divisor) of 5. Over the years we found that frequency compression greater than 5 was not effective. So, if we were listening to a 6000Hz tone, the lowest frequency the 6000Hz could be "moved" to would be 1200Hz. The digital frequency compression process used in Sonovation/AVR products is based on speech, not tones. Because speech sounds are not located at a single frequency, it is important to understand that the entire frequency region that comprises a voiceless sound gets compressed. All voiceless sounds will be compressed in the frequency domain by the same ratio.
Let me give you three brief examples:
Imagine then, we set our DFC ratio at 3.0. The results would yield:
This process allows the patient to gain access to voiceless sounds, which are lower in frequency than the original signal. Once the full spectrum of sounds are audible, patients learn to recognize the sounds in their speech and other people's speech. This leads to improved speech understanding, and in the case of children or individuals who have not adequately heard voiceless sounds in the past, improved speech production.
Adjustment to the new signal can be instant or it can take some time to re-learn what to do with the new information. Digital frequency compression, the way Sonovation/AVR has implemented it, lets audiologists get the most from their client's residual hearing. We can effectively move away from dead regions, reduce acoustic feedback, and give access to the critical speech sounds!
Beck: And you're not dumping the lowered voiceless sounds onto the voiced sounds and masking the voiced sounds - are you?
Davis: No. Voiced sounds are appropriately amplified based on gain and output requirements deemed necessary for the client's hearing loss between 250Hz - 2000Hz. When the phoneme analyzer in our digital frequency compression hearing instrument(s) recognize a voiceless sound is present, the sound gets compressed by the DFC ratio.
Beck: Wendy, this is really amazing technology. Who are the ideal candidates for this technology?
Davis: Thanks for asking. Candidates for Sonovation frequency compression will meet one or more of the following criteria;
Beck: Is the device available in all configurations, from CIC to BTE?
Davis: We have a completely digital line called ImpaCt Xp. It operates on a 17 band, 9 channel open platform DSP. Digital frequency compression is a programmable feature that can be enabled, or disabled on all products. We have CICs to BTEs, with and without directional microphones and a new BTE/FM called Logicom Xp. This product has an FM receiver integrated into the hearing aid case. We have just introduced a tiny, very sleek BTE with and without directional microphones called Nano Xp.
Beck: Very good. Thanks so much for your time this morning. This is amazing technology, and I want to thank you for explaining it to me!
Davis: Thank you too, Doug. It's always fun to speak with you too.
For more information about Sonovation hearing aids, click here.
Davis: Hi Doug, nice to be with you too.
Beck: If you don't mind, I'd like to start by having you explain about Sonovation's unique hearing aid product. It's actually very different from typical analog and digital hearing aids, and I wonder if you can give us an overview of the product itself, and also of Sonovation?
Davis: Sure Doug. Sonovation is a subsidiary of AVR Communications. Many people are used to looking for us under the name AVR Sonovation, but last year we started using our official U.S. name, Sonovation. Sonovation/AVR has always been associated with cutting edge technologies. Our approach has been to examine the needs of hearing aid users, assess what was currently available in the marketplace and develop products that address the unmet needs of the deaf and hard of hearing. Development has always been from the ground up, no restraints and no assumptions.
Beck: That's very exciting...and if I recall, your initial target market was the severe-to-profound hearing-impaired population - right?
Davis: Yes, that's where Sonovation/AVR started. We knew that audibility and ultimately recognition of voiceless sounds for people with severe-to-profound hearing loss was a real challenge. Even though hearing aids made environmental sounds and vowel (voiced) sounds louder, hearing aids generally didn't have a way to present the softest, highest frequency speech sounds, without saturating the hearing aid circuit, causing distortion, or creating acoustic feedback.
Beck: Sure, and that used to be a very common and enormously frustrating issue. By the time the high pitched sounds were loud enough to be perceived, they were too loud for the hearing aid, and the proximity of the microphone to the receiver, with adequate venting and enough high frequency amplification, and pretty soon we had feedback, or we turned down the amplification in the highs....which was a lot like a dog chasing it's tail....we just couldn't get there! Of course, 90 percent of the intelligibility of English is based on the high pitched sounds, so if you cannot hear those sounds, people sound like they're mumbling!
Davis: That's right, this was and continues to be a common dilemma, Sonovation/AVR had to "think outside the box." We knew the high frequencies were critical for improving speech understanding. We knew that voiceless sounds were the most difficult sounds to amplify through hearing aids without creating additional problems. We also knew that hearing is typically poorer in the high frequencies. To effectively address these issues, Sonovation/AVR
developed an algorithm to proportionally move high-frequency "voiceless" sounds
to lower-frequency regions. This is where patients typically have better hearing and this is where the hearing aid is more efficient at amplifying. This approach has been called "frequency transposition" but we now use the term "digital frequency compression."
Beck: Very interesting. And so sounds like /m/ and /ee/ and other voiced sounds are not affected while the voiceless sounds /sh/ and /s/ are compressed into the lower frequency range?
Davis: Exactly. Voiced sounds are amplified like all other hearing aids, while the voiceless sounds are moved to lower frequencies, where the patient can hear them. This process is done is real time, so there is virtually no delay. We're also following the dynamics of speech; vowel, consonant, vowel (VCV); consonant, vowel, consonant (CVC). Voiceless sounds are not "dumped" on top of voiced sounds. The process we are using is proportional so we are maintaining the spectral relationships within and between voiceless sounds. These relationships are what lets us identify that an "s" is different from an "sh" is different from a "t" and so on.
Beck: So if I had a sound at 3000 Hz, and I wanted to compress it down to 1500 or 750 Hz, would I just put in a mathematical number, like "compress this into half it's current value" or a quarter of it's value or whatever?
Davis: Sort of. We use the term "digital frequency compression" (DFC) ratio. This ratio is really a divisor from 1.25 to 5.0 in 0.25 steps.
Beck: So you could take a sound of 6000 Hz, and make that sound compress into 750 Hz without too much difficulty?
Davis: No, not really. The maximum we frequency compress is by a ratio (divisor) of 5. Over the years we found that frequency compression greater than 5 was not effective. So, if we were listening to a 6000Hz tone, the lowest frequency the 6000Hz could be "moved" to would be 1200Hz. The digital frequency compression process used in Sonovation/AVR products is based on speech, not tones. Because speech sounds are not located at a single frequency, it is important to understand that the entire frequency region that comprises a voiceless sound gets compressed. All voiceless sounds will be compressed in the frequency domain by the same ratio.
Let me give you three brief examples:
- An adult male produces an /sh/ sound. The energy of his /sh/ begins at
1500Hz and spreads up through 4000Hz. - An adult female produces an /sh/ sound. The energy of her /sh/ begins at
2500Hz and spreads up through 6000Hz. - A child produces an /sh/ sound. The energy of his/her /sh/ sound begins at
3000Hz and spreads up through 8000Hz.
Imagine then, we set our DFC ratio at 3.0. The results would yield:
- The man's /sh/ is compressed from 500Hz-1333Hz.
- The woman's /sh/ is compressed from 833Hz-2000Hz
- The child's /sh/ is compressed from 1000Hz - 2666Hz.
This process allows the patient to gain access to voiceless sounds, which are lower in frequency than the original signal. Once the full spectrum of sounds are audible, patients learn to recognize the sounds in their speech and other people's speech. This leads to improved speech understanding, and in the case of children or individuals who have not adequately heard voiceless sounds in the past, improved speech production.
Adjustment to the new signal can be instant or it can take some time to re-learn what to do with the new information. Digital frequency compression, the way Sonovation/AVR has implemented it, lets audiologists get the most from their client's residual hearing. We can effectively move away from dead regions, reduce acoustic feedback, and give access to the critical speech sounds!
Beck: And you're not dumping the lowered voiceless sounds onto the voiced sounds and masking the voiced sounds - are you?
Davis: No. Voiced sounds are appropriately amplified based on gain and output requirements deemed necessary for the client's hearing loss between 250Hz - 2000Hz. When the phoneme analyzer in our digital frequency compression hearing instrument(s) recognize a voiceless sound is present, the sound gets compressed by the DFC ratio.
Beck: Wendy, this is really amazing technology. Who are the ideal candidates for this technology?
Davis: Thanks for asking. Candidates for Sonovation frequency compression will meet one or more of the following criteria;
- People with hearing loss of 65 dB or poorer at 2000 Hz and higher.
- People with greater than 35db per octave slope.
- People with an inability to detect, discriminate or identify voiceless sounds.
- People with poor tonal perception.
- People with word recognition scores poorer than expected based on pure tone thresholds.
- Hard of hearing (HOH) infants and children in the critical speech and language developmental years.
- HOH individuals with poor speech production, i.e. slushy voiceless sounds.
- Medical conditions that are associated with tolerance problems or create high frequency hearing loss, such as chemotherapy patients.
Beck: Is the device available in all configurations, from CIC to BTE?
Davis: We have a completely digital line called ImpaCt Xp. It operates on a 17 band, 9 channel open platform DSP. Digital frequency compression is a programmable feature that can be enabled, or disabled on all products. We have CICs to BTEs, with and without directional microphones and a new BTE/FM called Logicom Xp. This product has an FM receiver integrated into the hearing aid case. We have just introduced a tiny, very sleek BTE with and without directional microphones called Nano Xp.
Beck: Very good. Thanks so much for your time this morning. This is amazing technology, and I want to thank you for explaining it to me!
Davis: Thank you too, Doug. It's always fun to speak with you too.
For more information about Sonovation hearing aids, click here.
