Oticon Frequency Lowering Access to high - frequency speech sounds with Speech Rescue technology

Frequency lowering is a well-known technology in hearing aids that shifts high-frequency sound to lower spectral regions. Research and clinical findings have revealed the challenges of developing a strategy that can provide access to high-frequency speech without introducing unacceptable amounts of low-frequency distortion. Oticon Speech Rescue, the latest frequency lowering processor on the market, is explicitly designed to transmit temporal features of high-frequency speech with minimal distortion of low-frequency spectral features. It is unique because it uses a multi-layered lowering technique that overlaps copied segments from a wide region in the high-frequency input in order to present the information in a narrow region in the low frequencies. Prescription of Speech Rescue is based on the principle of Maximum Audible Output Frequency (MAOF), which means that the lowered input will be at the border of the patient’s usable hearing. Here, we highlight the principles behind the Speech Rescue strategy and configurations, and describe the importance of providing a correct frequency lowering setting to optimise the benefit for the patient with a severe-to-profound hearing loss. E D I T O R S O F T H I S I S S U E Kamilla Angelo1, Joshua M. Alexander2, Thomas U. Christiansen1, Christian S. Simonsen1 & Claus F.C. Jespersgaard1. 1 Oticon A/S, Headquarters, Denmark 2 Dept. of Speech, Language, & Hearing Sciences, Purdue University Acknowledgements Thank you to Ryan McCreery for providing helpful insights and work with evaluating the Speech Rescue algorithm and settings. Thank you to Anne Specht Petersen and Maria Brorsson for running all clinical testing at the Oticon Headquarters, Denmark. Corresponding Author: If you have any questions to the content of the white paper please contact Kamilla Angelo, [email protected]. PAGE 2 WHITEPAPER 2015 – OTICON FREQUENCY LOWERING Introduction Perceptual importance of high-frequency energy – understanding the potential benefit A growing body of evidence shows that the high-frequency end of the speech spectrum plays a significant role in our perception of speech and voice quality, talker identification, speech source localisation and speech-in-noise performance (Monson et al. 2014). In particular, studies with hearing-impaired individuals have revealed that improved speech understanding in noise is possible when an effort is made to amplify the high frequencies (Hornsby et al., 2011; Levy et al., 2015; Plyler and Fleck, 2006; Turner and Henry, 2002). Furthermore, studies have shown that receiving insufficient audibility at high frequencies negatively affects the speech production, language development and word learning rate of hearing-impaired children (Pittman, 2008; Stelmachowicz et al., 2004). While access to high-frequency speech cues is not paramount to comprehension in quiet and favourable listening conditions, the additional information they provide becomes increasingly important for successful communication when listening conditions become complex. Highfrequency hearing loss can put individuals at a disadvantage since they do not have access to the full spectrum of speech cues that can facilitate communication in difficult acoustical environments. Consequently, these individuals may struggle to follow a conversation, may miss important information, and may become more easily fatigued. Depending on the degree and configuration of hearing loss, the greatest risk to speech perception is the inaudibility of consonants with significant high-frequency energy. For example, perception of fricative consonants such as “f”, “s” and “th”, which depends on frequencies above 4 kHz, may be completely missing from the incoming speech stream for individuals with severe-to-profound hearing loss. Depending on the linguistic context and background noise, information from the low frequencies may be inadequate for these individuals to fill in the missing content. At a minimum, they may have to apply additional cognitive resources when trying to perceive these sounds, thus making listening more effortful. Conversely, unlike an individual with normal hearing, they will not be able to rely on information gleamed from the high frequencies when information from the low frequencies is less reliable due to unfavourable signal-to-noise ratios, etc. “Frequency lowering can be viewed in terms of an improved audibility vs increased distortion Tradeoff” (Souza et al. 2013) Fortunately, with today’s hearing technology, these “out of reach” high-frequency sounds can now be restored within the usable bandwidth of hearing aid users. This can be done either by extending the bandwidth of conventional amplification for individuals with moderate highfrequency loss or by utilising signal processing strategies such as frequency lowering for individuals with more severe high-frequency loss. It is, however, important to realise that the challenge of using frequency lowering to provide access to high-frequency speech sounds is that it may come at the cost of distorting the natural frequency patterns contained in the low-frequency portion of the speech spectrum. Thus, pursuing high-frequency audibility by providing high-frequency gain with conventional amplification ought to precede the prescription of frequency lowering in modern hearing aids (AAA, 2013). However, in cases where it becomes necessary to employ frequency lowering technology, striking the right balance between improving access to high-frequency sounds and minimising low-frequency distortion is paramount to obtaining the optimal benefit for the individual patient (Alexander, 2013; Souza et al., 2013). Understanding the challenge Speech has evolved to be a remarkably robust signal that facilitates communication in the face of severe distortion. For example, the telephone transmits only a portion of the full speech spectrum (300 to 3300 Hz), yet intelligibility in quiet remains largely intact. Speech processed by cochlear implants, which transmit the slow modulations of speech in a sparse number of bands, can be largely understood in favourable listening conditions. However, communication begins to break down as sources of signal degradation accumulate. Sensorineural hearing loss is a major source of signal degradation that makes verbal communication quite frail when noise, reverberation and unpredictability in speech are added. Therefore, even though a hearing aid can present much greater spectral detail than a cochlear implant over a wider frequency range than a telephone, it can be very effortful and cognitively taxing to follow a conversation in noise. This fact, which has long been known among hearing aid developers, has inspired them to design signal processing strategies, such as noise reduction and directionality. More recently, there has been an awareness that information from the amplified speech signal can be made more robust by reintroducing the high-frequency cues that were given up on long ago due to limitations in receiver technology and the severity of loss in these regions. PAGE 3 WHITEPAPER 2015 – OTICON FREQUENCY LOWERING Frequency lowering technology different strategies Frequency lowering is the umbrella term for the signal processing in hearing aids that makes high-frequency sounds available at lower frequencies, where the patient has usable hearing. Today, frequency lowering is achieved in as many different ways as there are hearing aid manufacturers. However, conceptually, the different technologies use one of three basic techniques: compression, transposition, and composition (Fig. 1). With frequency compression (e.g. SoundRecover by Phonak) high frequencies are brought to lower frequencies by squeezing frequency content together in a smaller space. This is done for sounds above a selected start frequency, and distortion is thus introduced as the frequency spacing in a band is reduced to fit within the audible bandwidth of the patient. Depending on the position of the start frequency, the low-frequency spectra important for vowel identification are likely to be altered, at the risk of creating vowel confusion. Frequency transposition (e.g. Audibility Extender by Widex) captures a portion of the high-frequency spectrum and reproduces it at a lower spectral position, where it is mixed with the original signal. To avoid compressing the high frequencies only a small section is selected (Kuk et al., 2006; Kuk et al., 2009). Frequency composition is the latest technology on the market. It superimposes a highfrequency source band onto a low-frequency destination band, but it first divides the source band into 2 or 3 segments and then overlaps them in the destination band in order to present information from a wider input region in a narrower output region. For a comprehensive and recent review of the various strategies see Alexander (2013) or listen to the audiology online course #23437 “Individual variability in recognition of frequency-lowered speech” by the same author: http://www.audiologyonline. com/E/23437/113386/. Fig. 1 Frequency Lowering Strategies