For many hearing-impaired friends, hearing aids are indispensable for them to go on with their daily lives. As technologies advance, hearing aids will become smaller, more energy-efficient, and more acoustically legible. Artificial intelligence technologies will also improve hearing aids' effectiveness and functionality. This article will show you the technological development of hearing aids and the product characteristics of the Ezairo® 8300 audio processor introduced by onsemi.
New technology reduces hearing aid size and improves performance
About a decade ago, many people still wore large and obtrusive hearing aids, and users needed to constantly fiddling their positions, replace batteries, and complained that they still couldn’t hear clearly. In recent years, however, technological advances have transformed these devices from discouraging sources into advanced audio devices connected in modern IoT-based ecosystems, thus providing users with significant benefits and excellent audio performance.
The first step in the evolution of hearing aids is to perform frequency-based nonlinear amplification, with each band being amplified according to the level of energy contained in the band. As a result, loud noise is weakened and quiet sounds are amplified, which will greatly enhance the experience of hearing aid users.
This progress is driven mainly by the emergence of digital signal processing (DSP), which gives hearing aids the granularity to customize and amplify different frequencies, allowing them to be tuned and optimized for users, especially when their hearing degraded over time. With DSP, audio signals can be handled in a completely new way with highly selective digital filters that isolate frequencies and thus also handle sound types.
Hearing aids are also growing intelligently as a result of DSP, enabling them to adapt to their environment. For example, hearing aids can infer that the wearer is in a noisy environment, such as crowds, mobile vehicles, or large public places. The hearing aid will be reconfigured to the audio environment based on real-time control and adjustment algorithms for digital filters to provide the best possible audio experience in all scenarios.
In addition, the rapid development of wireless communication technologies, such as the personal area network provided by Bluetooth® Low Energy (BLE) connections, has also had a positive impact on hearing aid technology. Modern networked hearing aids can be controlled by smartphone apps and can be used for phone/video calls or for listening to smartphones, tablets, or any other form of audio for any other BLE-enabled device, including voice assistants.
Hearing aids add more health features through artificial intelligence
On the other hand, technologies such as artificial intelligence (AI) and machine learning are beginning to enter hearing aids to further expand their functionality and performance. Many believe that this technology will have an unprecedented impact. Looking ahead, the human ear also offers considerable opportunities for sensing. In the near future, many hearing aids will include sensors for medical purposes, such as heart rate monitoring, blood oxygenation, blood glucose metering, and fall detection. In addition, the same sensors can be used for consumer-oriented health functions such as step-counting.
As hearing aids provide more value-added functions, such as streaming audio from personal devices, the distinction between them and emerging product groups known as hearables (over-the-counter hearing aids, earbuds, etc.) is becoming smaller. Audiovisual technology opens up many potential sources of revenue, including real-time translation, listening, and navigation services. In fact, anything that can provide a voice through BLE connections can be provided through listening programs, including modern hearing aids.
Without significant advances in semiconductor manufacturing technology in recent decades, as well as other significant advances in DSP and BLE technologies, there would be no modern hearing aids. Tracking processors over time is usually based on the number of integrated transistors. The first personal computer (PC) processors had tens of thousands of transistors manufactured using semiconductor processes and featured sizes within a few micron-meters. Today, these processors have tens of billions of transistors, and their technology nodes are thousands of times smaller. This level of integration benefits not only the PC industry, but also the entire semiconductor industry with evolving technologies that enable all vertical market manufacturers to move forward.
Specifically, the market for hearing aids has benefited from a multi-core parallel processing architecture with exponential improvements in processing power, minimization of clock cycles, and a significant reduction in power consumption. Such cores can process highly sophisticated algorithms, including AI.
Audio processors are an excellent choice for high-end wireless audio applications
Many modern audio products will benefit from the latest devices in onsemi's DSP-based EZAIRO® series audio processor solution. The Ezairo® 8300 is designed and developed specifically for hearing aids and hearables devices and has six processing cores, each of which optimizes performance for this application area. This includes three DSP cores, a microcontroller, and two hardware accelerators, one of which is a neural network accelerator designed to execute AI and machine learning functions efficiently without requiring additional processor capabilities.
The Ezairo® 8300 consists of six programmable or semi-programmable cores that provide a high degree of parallelism and flexibility, including CFX, an open-programmable, dual-Harvard 24-bit DSP that supports any type of audio signal processing, and an Arm® Cortex®-M3, a 32-bit RISC processor that supports general processing and interfaces with external components, combines HEAR configurable accelerator core, optimized for pre−programmable features that are often required for audio signal processing, and includes filter engine that allow time domain filtering and support ultra-low−delay audio path. In addition, the LPDSP32 is an open-programmable, dual-Harvard 32-bit DSP, and a neural network accelerator that allows the Ezairo® 8300 to perform neural network computing efficiently and flexibly.
The Ezairo® 8300 includes four ADCs with signal detection modes and two direct digital output drivers with high quality and ultra-low power consumption. When used in combined with non-volatile memory and wireless transceivers, the Ezairo® 8300 also includes the peripherals and interfaces needed to make it a complete hardware platform.
The Ezairo® 8300 has a dynamic range of 108 dB and a high-fidelity audio system with a sampling frequency of up to 48 kHz that supports dynamic selectable system clock speeds from 1.28 MHz to 61.44 MHz, and is suitable for versatile memory architectures with 1433 kB memory.
The Ezairo® 8300 provides high-fidelity operations at very low power consumption, multiple interfaces, flexibility to maximize power consumption performance, and support for BLE connectivity, making the Ezairo® 8300 an excellent option for high-end wireless audio applications.
Add artificial intelligence to audio products using neural network accelerators
The Ezairo® 8300 features multiple DSPs for signal processing, improves the definition of the received audio, further analyzes the audio signal, and has a built-in Neural Network Accelerator (NNA), which is a configurable hardware accelerator block dedicated to supporting energy-efficient neural networks. It allows complex neural networks to operate in an energy-efficient manner. Accelerators can perform single-layer fully populated or sparsely populated neural networks in a single task without any processor intervention and can support up to 1023 inputs and 1023 outputs layers.
The NNA contains 16 multipliers, 16 accumulators, 16 input registers, and 16 coefficient registers. It includes inputs and coefficient “fetchers” that, once configured, automatically manage data and coefficient memory access, including support for coefficient compression/decompression and pruning, help minimize the number of coefficients required, reduce the burden on system processors, and improve the performance of artificial intelligence processing.
The Ezairo® 8300 can be applied to high-resolution stereo audio streaming, intensive audio processing, ULP companion chip for Bluetooth Audio, and common end products include wireless-enabled hearing aids, earbuds/hearables, personal sound amplification products (PSAPs), over-the-counter (OTC) hearing aids, headphones, etc.
onsemi also provides an open-programmable evaluation and development kit (EDK) for the Ezairo® 8300 and is available as a subscription with 2, 5, or 10 user seats.
Conclusion
With the rapid development of technology, hearing aids have been used not only for hearing aid functions, but also benefit more hearing-impaired people by combining audio playback earphone functions with sensor health functions. Apart from hearing aids, onsemi's Ezairo® 8300 audio processor is an ideal choice for various headphones and audio playback devices. Its built-in DSPs and NNAs will enhance the added value of audio products, differentiate products in the market, and increase product competitiveness. It will be worthy of your further understanding and adoption.