In a previous paper,1 we provided a tutorial on the principles of digital audio codec and outlined some of the components in a wireless hearing aid. In this paper, we will discuss the important factors to consider in a digital wireless hearing aid and how those considerations are integrated into the design of the digital wireless platform used by WidexLink, a new technology that enables the C-Integrated Signal Processing (C-ISP) used in the CLEAR440 hearing aids.
Three Important Factors in the Design of a Digital Wireless System
Because WidexLink is designed for hearing aid applications, the description here is applicable for both audio and data exchange between the two partner hearing aids and external devices. In particular, a technical description on how WidexLink considers the important design criteria of 1) excellent sound quality, 2) robust communication, and 3) efficient battery life is provided. The audiological applications of such coordinated data exchange in a binaural fitting (InterEar) will be described in Part 3.
#1) Sound Quality
Hearing aid. In previous papers, we have considered the factors that affect the sound quality of a hearing aid.2,3 A digital wireless hearing aid that possesses excellent sound quality obviously needs to have good sound quality from both the hearing aid and the wireless codec system. As a quick review, a low noise floor, a high input dynamic range, a high sampling rate, a high bit resolution, a broad bandwidth, an adequate MPO, and advanced compression algorithms at the various stages of the signal flow are key components to ensure a good sound quality. The use of slow-acting compression as the primary mode of signal processing also preserves the natural dynamics of the input sounds and retains all the nuances of the inputs.
Codec. The sound quality provided by the wireless codec is highly dependent on the method that is used to compress the data, how the data are transmitted, and the tolerance of the audio codec toward errors that may occur during the transmission process. As indicated in the previous paper, a codec compresses the audio data to reduce the size of the data set so that the same information can be transmitted more efficiently without appreciable delays and artifacts. Thus, an effective codec algorithm would more likely retain the full range of sounds without any loss of perceptually significant information. This ensures superior sound quality of the transmitted sounds.
Bit rate. The channel capacity of a wireless system also plays an important role in the sound quality of the transmitted sounds. Typically, a system that has a higher capacity or bit rate would have a better sound quality because more information can be transmitted per unit time.
However, a higher bit rate does not guarantee a good sound quality. This is because the original uncompressed signal may be of poor quality or the codec algorithm may not have preserved the important cues critical for good sound quality (eg, introducing more noise and missing important details). In both cases, a high transmission bit rate cannot compensate for the poor quality of the original signals.
Coding delay. It is inevitable that, during transmission, electromagnetic interference occurs from other wireless sources, corrupting the integrity of the transmitted signal. How the codec handles errors from interference during transmission could delay the transmission and affect sound quality.
Coding delay may be inconsequential when the listener simply listens to transmitted sounds (such as playback from an MP3 player). On the other hand, when the listener listens to the transmitted sounds along with the direct sound (or amplified direct sound, such as in an open or vented fitting), or listens to the transmitted sound while watching its source (such as a TV), the effects of the delay can be consequential. A small delay that is less than 10 ms may not be perceptible; a delay of 30-50 ms could lead to a “metallic” sound quality and the perception of an “echo.”4 A delay of over 100 ms could lead to a dys-synchrony between the visual and auditory information.
A low delay is a prerequisite to ensure good sound quality in a digital wireless communication system that is integrated into a hearing aid.
#2) Robust Communication
Long-range versus short-range. Protecting the audio data from the effects of interfering electromagnetic noise in the environment is a real and serious design issue. A system that is robust ensures continuity of the data transmission and goodness of the audio data; a non-robust system can experience drop-outs of the signals and/or poor sound quality from interference.
Factors that affect the robustness of the communication system include the distance of the transmission, the method that is used for the transmission, and how the system handles errors. As indicated earlier, audio data can be transmitted over a short range or over a long range. The closer the transmitter is to the receiver, the stronger the received signal and the lower the likelihood of transmission errors.
Typically, short-range transmissions are less likely to interfere with and less susceptible to interference from other short-range transmissions. Conversely, long-range transmissions are more susceptible to interference.
Codec. A codec has several ways to ensure the accuracy and robustness of the data transmission. This is done in the encoding stage through the channel coding process where details of the data set, along with error correction codes, are provided. A codec that has a robust way of encoding is less susceptible to errors.
Transceiver. A transceiver that has a reliable way of detecting and identifying the transmitted bits makes fewer errors when decoding the transmitted signals. Indeed, minimizing the potential sources of errors during the transmission stage and correcting them during the decoding stage is the most critical step in ensuring a robust system.
Previously, we indicated that digital wireless transmission is less susceptible to interference than analog wireless transmission. Most power-limited wireless digital transmission systems use Frequency Shift Keying (FSK) to transmit digital signals. How this radio technology identifies the coded audio in the presence of noise is important to ensure the integrity of the signal. And, just as important, how it handles errors affects the outcome.
For example, many Bluetooth systems handle errors by requiring the encoder to re-send the data. This means the decoder will not use the incorrect data but instead waits for the correct data to be sent. This is one reason why a Bluetooth system can experience a delay as long as 150 ms. Other systems, like the earlier version of the Digital Audio Broadcast (DAB) system, do not permit data re-send and have limited ability to correct errors. This leads to drop-outs of sounds and periods of silence until the correct data are received.
It is desirable to have a system that does not encounter any errors during the transmission process. But, since that is not possible, a system that can handle errors without drop-outs, severe artifacts, and/or delays is desirable.
#3) Efficient Current Drain
Many of the issues with sound quality and robustness are non-issues if the power supply to the wireless hearing aid system is unlimited. This is obviously neither practical nor possible in hearing aid applications because of size constraints. In real life, a design that uses the least amount of power from the hearing aid while achieving an acceptable level of robustness and good sound quality is adopted. Several factors that affect the current drain include the complexity of the codec, the transmission bit rate, the transmission power, and the complexity of the transceiver.
A codec that does minimal data compression (ie, no data reduction) requires a lower current drain than one that does a lot of data compression. But little data reduction also means a larger data set to be transmitted—which ultimately increases current drain. One way to have complex processing while keeping an acceptable current drain on the hearing aid is to keep the coding complexity in the audio encoder and have minimal complexity and sophisticated processing in the decoder. This is because the encoder (for external sources) is typically outside of the hearing aid where a larger battery or direct power from the line supply is possible. By keeping most of the processes that require very high computing power outside the hearing aid, it is possible to optimize the life of the hearing aid battery while maintaining the complexity of the processing.
In real life, a compromise between the amount of data compression (to reduce data size), the transmission bit rate and/or bandwidth, sound quality, and power consumption needs to be made. The degree and form of compromise varies among designers. Thus, no two wireless platforms are identical in their effectiveness. The quality of each platform must be evaluated based on the “trade-offs” that the system makes to achieve wireless connectivity and how those trade-offs may (or may not) affect performance.
WidexLink Digital Wireless Platform
The WidexLink digital wireless platform was developed to provide customized wireless communication solutions with a design imperative to preserve the high sound quality provided by CLEAR440 hearing aids.
WidexLink covers both short-range and long-range communications, and allows coordination and synchronization of hearing aid parameters between the left and right hearing aids. It also allows optional user control of selected hearing aid features through the use of the new versatile remote control. Most notably, it allows high-fidelity transmission of audio signals between hearing aids and from external sound sources (eg, TV, cell phone, MP3, telecoil, etc) to the hearing aids.
The system is developed on a proprietary technology platform based on Widex’s vision of future needs. The criteria for this platform include a very robust audio codec, a superior sound quality with a low delay, and a reasonable current drain.
Audio codec. Figure 1 shows a schematic block diagram of the components of the WidexLink as it is integrated within the CLEAR hearing aid. Although all wireless systems include an encoding (or coding) stage, a decoding stage, and a transceiver that transmits and receives the signals, the encoding stage is not included in Figure 1 because it is external to the hearing aid. The general utility and a description of each stage have been described in the previous paper.1
FIGURE 1. Hearing aid with digital wireless communication technology and C- ISP processor.
The core of the proprietary WidexLink audio codec is inspired by an adaptive signal model called Analysis by Synthesis. This is a highly complex and sophisticated method of coding and decoding audio signals within a closed (or bit-true) system. In short, instead of transmitting the audio signal in its original form, this approach analyzes the content of the audio signal at a very fast rate, then sends the result of the analysis to the receiver so it may recreate or synthesize the original signal.
An analogy to the Analysis by Synthesis model may be made to shipping cakes. Imagine that you are the owner of a bakery and your cakes are very popular. People all over the world order cakes from you. For the average customers who order a limited number of cakes on an infrequent basis, you ship the cakes directly to them. In the process, you pack the cakes in a sturdy container so they may not be dented (or worse yet, squashed). You ship it by express mail so they arrive fresh. In some situations, accidents happen to the cakes so they may be delayed or damaged during the shipping process. Obviously, you will have to add the costs of shipping, packing, and potential damages to the cost of the cakes.
Imagine now you have a contract to supply cakes to a large specialty grocery store, which also has an in-house bakery. In this case, shipping finished cakes to the grocery store may not be the most efficient use of resources. Because the specialty store also has a bakery, a more efficient way to ship cakes to the stores is to ship your recipes with detailed instructions and ready-made pre-mix cakes to the specialty store. When the store receives the cake mix, its bakers can bake the cakes themselves using your ingredients while following your instructions. This way, the cost involved in shipping the cakes is significantly reduced. The number of bags of cake mix sent can be increased. Fewer special precautions are necessary when packing the cake mix. The number of accidents that can happen to damage the cakes is reduced. And, most importantly, the cakes will be available at the specialty store fresh and according to your high quality standard. Obviously, this approach is only possible when the receiver (ie, specialty grocery store) also has bakers who know how to bake the cakes with the special cake mix that you sent.
The act of packing and shipping your cakes is analogous to the task faced by engineers when designing wireless transmission systems. In this case, engineers have to consider how to transmit the most audio signals in the most efficient and error-free manner.
In the Analysis by Synthesis model, audio inputs are analyzed first to understand the component sounds that make up the audio signal. Because there are only a finite number of sound components, each component can be matched to a special code. This code, rather than the actual sound or sound components, is transmitted wirelessly. Because both the transmitter and the receiver (transceiver) are designed by Widex, the special code that is received can be easily translated into the appropriate sound components. Afterward, the sound components are synthesized to recover the original audio signal. This is only possible for a bit-true system used in the WidexLink.
There are several important benefits from using this approach. First, it dramatically reduces the number of transmitted bits and saves on the current consumption. Second, the same number of bits carried by the system will have more details or information of the audio data than the same number of bits carried by another codec that uses a simpler method of coding and decoding. And third, since only the special codes are transmitted, its interception by an unintended receiver will not be meaningful because it cannot be decoded back. Thus, data security is enhanced. Kuk et al1 described some simpler methods of codec in the primer paper.
In WidexLink, the transmitted signals are sampled at 25 kHz to result in a realizable transmission bandwidth from 100 Hz to 11.2 kHz—which is even broader than the broad bandwidth (10 kHz) of the CLEAR440 ClearBand model (CL4-m-CB) when it is in the typical mode of operation (ie, microphone mode).
As is common in digitization, audio codec introduces quantization noise at the output. In many codec, the level of the noise is independent of the level of the signal (ie, it is fixed). As a result, the signal-to-noise ratio (SNR) decreases when signal level decreases. In the WidexLink codec, the quantization noise level adapts to the level of the input signal. As the signal level decreases, the quantization noise decreases in order to maintain a fixed SNR. This SNR level is set to be inaudible to the user, while still allowing for a significant amount of digital audio compression.
Channel coding. In designing the WidexLink, special attention has been paid to ensure a robust system that can tolerate electromagnetic interference during transmission. When interference occurs, the content of the transmitted signals may change. The difference between the “intended-for-transmission” signals and the transmitted signals is called an error. Channel-coding algorithms provide a method to ensure and check the accuracy of the received data. They also include additional codes that specify how errors are handled, and these error correction codes are different among manufacturers. The codes can range from simply identifying an error to those that can anticipate the errors and correct for them—an approach called Forward Error Correction (FEC).
The error correction method in the WidexLink is a FEC type that is capable of anticipating what types of errors will likely occur. As a consequence, a high proportion of the transmitted 212 kbits are designated for error correction purposes so when errors occur, they are corrected immediately. As a result, audio information is continuously passed to the user to ensure minimum drop-out of sounds and that the transmitted sounds are of good quality. This differs from simpler digital audio transmission systems where corrupted audio samples are either removed (resulting in a complete drop-out of the audio) or the data are re-transmitted (which delay the transmission). The error correction codes in the WidexLink are one of the many lines of defense in ensuring exceptional sound quality and robustness of the wireless communication system.
Before sending the encoded digital audio, the channel encoder adds the FEC, the address of the receiver, command data, and a data type identification code that specifies which data are instructions, which are audio data, and which are error correction codes so the receiver knows how to process the transmitted data. This also ensures correct transmission of the audio to the intended hearing aid, thereby preventing eavesdropping. This step also eliminates potential interference from other hearing aids in the surroundings and ensures security of the data as well. The data type ID code lets the receiving hearing aid know if the audio data that it is receiving is intended for mono or stereo playback. If the data type ID has changed, the hearing aid will check the command data to determine in what format the audio data are being transmitted and make appropriate corrections.
The channel decoder of the receiving hearing aid can look at the added FEC information and compare it with the received audio data. If the audio data agree with the FEC data that are sent, then the receiving hearing aid can be confident that the received audio data are indeed correct and error-free. If the audio data do not agree with the FEC, the receiving hearing aid will register that an error has occurred and initiate the correction process.
In addition to a way of reducing the size of the load and the meticulous use of forward error correction to ensure robustness, an accurate means to detect and identify the transmitted bits is also important to ensure robust transmission. This is achieved through the use of the patented, robust WidexLink transceiver technology. Of note is the use of a new oversampling method for receiving the wirelessly transmitted signal that is modulated with the Frequency Shift Keying (FSK) technique.
Figure 2 shows a block diagram of the functional components in the WidexLink transceiver. In order to use the same antenna for transmitting and receiving the FSK modulated signals, a receiver/transmitter switch is used. As a standard means, two radio frequencies are used to transmit the digital “1” and “0.” The transmitted “1” and “0” are transformed into a voltage with a rising slope and a falling slope, respectively.
FIGURE 2. The WidexLink transceiver block diagram.
The received FSK signal is first demodulated. The task for the decoder is to detect if the voltage slopes are rising or falling. Conventional FSK detectors sample twice per slope to determine if it is rising or falling (Figure 3). In general, this is sufficient for a correct identification when no interference is present. However, if the FSK signal is contaminated by interference, it might be difficult to detect the slopes correctly. Figure 3 shows that, when interference is present (red curve), the detected signal (orange line) may suggest a bit “0” instead of a bit “1” because of the estimated falling slope when it is rising (green in Figure 3).
The proprietary WidexLink transceiver system utilizes an “oversampling” technique where multiple (5) samples are taken per slope to determine if the slope is rising or falling (Figure 4). This allows greater certainty in the identity of the transmitted bit. Thus, the WidexLink transceiver is designed to be more accurate and sensitive than traditional FSK transceivers, especially when interference is present. The end result is a more robust communciation system with consistently exceptional sound quality.
The previous discussion provided a general description of the considerations behind the WidexLink codec used in the CLEAR440 hearing aid. To take advantage of the different properties of radio waves, WidexLink is utilized in both short-range and long-range applications. They are briefly described here, but their audiological ramifications will be further explained in the next paper.
Short-range between CLEAR440 hearing aids. A short-range communication is used to exchange synchronization and coordination data between hearing aids at 21 times per second. In addition, audio is exchanged between the two hearing aids, as well as between the hearing aids and streamer devices. This type of communication uses magnetic induction with a carrier frequency in the mega-hertz (MHz) radio frequency range. This uses much less power (approximately 50 times less) than long-range communications such as Bluetooth and WiFi. Although short-range radio technology can be used up to 1 meter, the range of transmission between hearing aids is approximately 30 centimeters (about 1 foot) using standard hearing aid batteries. The sensitivity is highly dependent on the transmitting and receiving antennae having the same orientation.
Short-range between assistive listening devices and CLEAR440. Short-range communication is also used between the remote control (RC-DEX) and the hearing aids. In addition, the use of optional gateway assistive listening devices (ALDs) allows routing the audio signal received from the external devices, such as a TV or cell phone, to the hearing aids. Because of the short distance, this gateway device can make use of the short-range inductive transmission technology to transmit the commands from the RC-DEX or the audio data from the external devices to the hearing aids.
Proprietary long-range between TV and Streamer (TV-DEX). The WidexLink platform is also implemented on ALDs that use a proprietary long-range strategy to transmit the audio data. One is to transmit data from the TV via the TV-base transmitter to the body-worn TV-controller (or streamer). The TV-base provides two mini-jack inputs that allow a TV and one other audio device to input into the TV-base. It uses an electromagnetic wave at a carrier frequency of 2.4 GHz (similar to Bluetooth), and the TV-DEX transmits audio data in a stereo format to the hearing aids.
The current consumption of this long-range transmission is higher than would be normally acceptable for use in a hearing aid. Because the transmission originates from the TV-base, which is connected to a standard wall outlet for power, current drain on the hearing aid battery is not a problem. Additionally, due to the efficiency of the codec, the audio delay between the TV and the hearing aid is less than 10 ms (end to end).
Bluetooth between mobile phone and streamer (M-DEX). Because Bluetooth wireless connectivity is an industry standard for cellular telephones, Bluetooth is used for communication between a cell phone and a cell phone ALD, called the M-DEX. The higher audio delay of Bluetooth (between the M-DEX and the phone) is still present in this connection. But, since it is only used in the auditory mode (listening to the transmitted cell phone sounds alone), the effect of the delay is not a problem in this application. There is a microphone on the M-DEX that can be used to pick up the hearing aid user’s voice. A lanyard is available for the wearer to hang the M-DEX around the neck. Thus, the wearer can use the M-DEX in conjunction with the CLEAR440 and the cellular phone in a hands-free mode.
The newly developed WidexLink wireless communication technology is a proprietary digital radio-frequency transmission technology that is designed to provide high audio quality and efficiency. WidexLink offers new possibilities for extended bandwidth audio streaming from external devices (DEX) to the hearing aids, and the continuous exchange of coordination and synchronization data between hearing aids.
The unique digital wireless connection is designed to offer an unparalleled short delay when transmitting audio. This ensures low distortion and echo-free audio quality when using hearing aid microphones together with direct audio transmission to the hearing aids through optional ALDs.
- Kuk F, Crose B, Korhonen P, Kyhn T, Mørkebjerg M, Rank M, Kidmose P, Jensen M, Larsen S, Ungstrup M. Digital wireless hearing aids: Part 1: A primer. Hearing Review. 2010;17(3):54-67. Accessed May 1, 2011.
- Kuk F, Korhonen P, Baekgaard L, Jessen A. MPO: A forgotten parameter in hearing aid fitting. Hearing Review. 2008;15(6):34-40. Accessed May 1, 2011.
- Kuk F, Jessen A, Baekgaard L. Ensuring high fidelity in hearing aid sound processing. Hearing Review. 2009;16(3):34-43. Accessed May 1, 2011.
- Stone M, Moore B. Tolerable hearing aid delays. I. Estimation of limits imposed by the auditory path alone using simulated hearing losses. Ear Hear. 1999;20(3):182-192.