Digital signal processing fundamentals

Modern hearing aids are sophisticated digital signal processors. The core signal processing chain converts acoustic sound to a digital signal via a microphone and analog-to-digital converter, processes the digital signal through a series of algorithms, converts back to analog, and delivers amplified sound to the ear via a receiver (speaker). The entire process must occur within a processing delay short enough that the processed sound does not perceptibly lag behind the direct sound reaching the ear — typically under 10 milliseconds, with premium devices achieving 5ms or less.

Key DSP functions in modern hearing aids include:

• Multi-channel compression: The frequency range is divided into multiple channels (typically 16–64), with independent compression applied in each channel. This allows the hearing aid to provide more gain in frequency regions where the patient has greater loss, without over-amplifying regions with less loss.
• Noise reduction: Algorithms that detect and attenuate non-speech noise signals. Modern noise reduction systems use modulation-based detection (speech has characteristic amplitude modulation patterns that differ from steady-state noise) and can reduce noise without significantly degrading speech.
• Feedback management: Algorithms that detect and cancel acoustic feedback (whistling) before it becomes audible. Phase cancellation and notch filtering are common approaches. Effective feedback management allows higher gain without feedback, particularly important for severe loss fittings.
• Directional microphone processing: Using two or more microphones to create a directional polar pattern that attenuates sound from behind and the sides while preserving sound from in front. Adaptive directionality adjusts the null direction in real time to track the dominant noise source.

AI and deep neural networks in hearing aids

The integration of artificial intelligence — specifically deep neural networks (DNNs) — into hearing aid signal processing represents the most significant technological shift in the field in the past decade. DNNs trained on large datasets of real-world sound scenes can classify acoustic environments and optimize processing parameters with a sophistication that rule-based algorithms cannot match.

Current clinical implementations include:

• Oticon's Deep Neural Network (DNN): Trained on 12 million real-life sound scenes, Oticon's DNN drives the MoreSound Intelligence processing in the More, Real, and Intent platforms. Rather than suppressing non-speech sounds, the DNN preserves a balanced sound scene that supports the brain's natural auditory processing — the BrainHearing philosophy.
• Phonak Sphere Integrated Xperience: The Sphere platform (2024) introduced a dedicated AI chip separate from the main audio processor. The AI chip handles environment classification and processing optimization without competing with audio processing for computational resources.
• Starkey Genesis AI: A dedicated AI processor handles health monitoring (fall detection, activity tracking) alongside sound processing, enabling simultaneous health and audio functions without battery life compromise.
• Widex SoundSense Learn: Uses machine learning to personalize sound preferences through A/B comparisons in real-world environments. The system learns individual preferences over time and applies them automatically.

The clinical implications of AI processing are significant. AI-driven systems can adapt to acoustic environments that rule-based systems cannot classify, produce more natural sound in complex environments, and — in the case of Oticon's DNN — represent a fundamentally different philosophy about what hearing aids should do. Understanding these differences is essential for counseling patients about technology choices.

Bluetooth and wireless connectivity

Wireless connectivity has transformed hearing aid functionality, enabling direct audio streaming, remote control, remote fine-tuning, and integration with smartphones and other devices. The connectivity landscape is currently in transition:

• Made for iPhone (MFi): Apple's proprietary protocol for direct hearing aid streaming, introduced by ReSound in 2014. MFi provides low-latency, high-quality audio streaming from iPhones and iPads. Most major manufacturers offer MFi-compatible devices.
• ASHA (Audio Streaming for Hearing Aids): Google's Android protocol for direct hearing aid streaming, introduced in Android 10. ASHA streaming quality and reliability have improved significantly but historically lagged behind MFi.
• Bluetooth Classic: Used for audio streaming via an intermediary device (streamer) in older hearing aid platforms. Higher power consumption than LE Audio.
• Bluetooth LE Audio: The next-generation Bluetooth standard, based on the LC3 codec, that provides lower power consumption, better audio quality, and new capabilities including Auracast broadcast audio. ReSound Nexia was among the first hearing aids to support LE Audio. LE Audio will eventually replace both MFi and ASHA as the universal connectivity standard.
• Auracast: A Bluetooth LE Audio feature that enables broadcast audio — a single transmitter can stream to an unlimited number of receivers. Auracast-equipped venues (theaters, airports, lecture halls) will be able to stream audio directly to compatible hearing aids, eliminating the need for hearing loops or FM systems.

Rechargeable technology

Rechargeable hearing aids have become the dominant form factor in the premium and mid-range segments. Current rechargeable technologies include:

• Lithium-ion: The most common rechargeable technology. Provides 16–24 hours of use per charge (including streaming), charges fully in 3–4 hours, and supports fast charging (3 hours of use from 30 minutes of charging). Li-ion batteries are sealed in the hearing aid and not user-replaceable.
• Silver-zinc: Used in some custom ITE styles. Rechargeable but with a shorter battery lifespan than Li-ion (typically 6 months before replacement). Allows rechargeable technology in smaller form factors.
• Disposable zinc-air: Still preferred by some patients for its user-replaceability and availability. Size 312 (brown tab) is most common for RIC styles; size 13 (orange tab) for BTE; size 10 (yellow tab) for CIC/IIC.

Emerging technology: over-the-counter and self-fitting

The FDA's 2022 OTC hearing aid rule created a new regulatory category for self-fitting hearing aids for adults with mild to moderate hearing loss. OTC devices use smartphone apps to guide users through a self-fitting process — typically a brief in-app hearing check followed by automatic programming. The clinical implications for audiologists are significant:

• OTC devices use simplified fitting algorithms that cannot match the precision of REM-verified prescriptive fitting
• Self-fitting apps vary widely in quality — some (Sony Hearing Fit, Jabra Enhance app) are well-designed; others provide minimal guidance
• OTC devices are appropriate for mild to moderate loss only; patients with severe loss, asymmetric loss, or complex audiometric configurations require professional fitting
• Many patients will try OTC aids before seeking professional care — understanding OTC technology helps audiologists counsel these patients effectively

Implantable and bone-anchored hearing devices

For patients who cannot benefit from conventional hearing aids, implantable options include:

• Cochlear implants (CI): For severe to profound sensorineural hearing loss. The electrode array is surgically implanted in the cochlea; the external processor converts sound to electrical signals delivered to the auditory nerve. CI candidacy criteria have expanded significantly — many patients with aidable hearing in one ear are now candidates.
• Bone-anchored hearing aids (BAHA): For conductive hearing loss, mixed hearing loss, or single-sided deafness. A titanium implant is placed in the skull; the external processor transmits sound vibrations through bone to the cochlea. Softband and adhesive options are available for patients who are not surgical candidates.
• Middle ear implants: For patients with sensorineural or mixed loss who cannot wear conventional hearing aids. The Vibrant Soundbridge (MED-EL) and Esteem (Envoy Medical) are examples.