Audio processing is the manipulation of sound signals to enhance, change, or analyze them. It covers everything from adjusting the bass on your car stereo to the noise cancellation in your headphones to the compression algorithms that make streaming music possible. At its core, audio processing takes a sound signal, applies some transformation to it, and outputs a modified version.
Analog vs. Digital Processing
Sound in the real world is analog: a continuous wave of air pressure changes. Early audio processing was entirely analog, using physical circuits with capacitors and resistors to shape sound. Turning the tone knob on a vintage guitar amp is analog processing in action.
Digital audio processing, which dominates today, works differently. It requires converting that continuous sound wave into a stream of numbers a computer can manipulate. This happens through two approximations. First, the sound’s amplitude (its loudness at any given moment) gets rounded to the nearest available value, a process called quantization. Second, the continuous wave gets sampled at regular intervals rather than captured in one unbroken flow. A CD-quality recording samples the sound 44,100 times per second (44.1 kHz) using 16-bit depth, giving about 65,000 possible loudness levels per sample. Professional music production typically uses 24-bit depth at 44.1 or 48 kHz, which provides over 16 million possible values per sample and much finer detail.
Once in digital form, the sound can be manipulated mathematically in ways that would be impossible with physical circuits. When you’re done processing, a digital-to-analog converter turns the numbers back into an electrical signal that drives your speakers or headphones.
Frequency Shaping and Equalization
The most familiar type of audio processing is equalization, or EQ. This is what you’re using when you boost the bass or reduce the treble. EQ works by filtering specific frequency ranges, making some louder and others quieter.
A few basic filter types make this possible. A high-pass filter lets higher frequencies through while cutting low ones (useful for removing rumble from a recording). A low-pass filter does the opposite, passing bass while cutting treble. Shelving filters boost or reduce everything above or below a set frequency by a fixed amount. These are what the bass and treble knobs on most stereos use.
Parametric equalizers offer more precise control. They let you select a specific center frequency, choose how narrow or wide the affected range is, and then boost or cut that range by a set amount. A music producer might use a parametric EQ to cut a narrow band around 3,000 Hz to reduce harshness in a vocal recording, or boost a wide band around 100 Hz to add warmth to a bass guitar.
Dynamic Range Control
Dynamic range is the gap between the quietest and loudest parts of an audio signal. Compression narrows that gap by automatically reducing the volume of sounds that exceed a set threshold. This is why a pop song on the radio sounds consistently loud rather than swinging between whisper-quiet verses and blasting choruses. A limiter is an extreme version of compression that acts as a hard ceiling, preventing the signal from ever exceeding a certain level.
These tools have parameters that shape how they respond. The threshold sets the volume level where processing kicks in. The ratio determines how aggressively loud sounds get turned down. Attack controls how quickly the processor reacts to a sudden loud sound, while release controls how quickly it stops compressing after the sound drops back below the threshold. Compressors typically react faster to sudden loud sounds than they do to fading ones, which keeps the processing from sounding unnatural.
Noise Cancellation
Active noise cancellation in headphones is one of the most common real-world applications of audio processing. It works by exploiting a physical property of sound waves: when two identical waves are perfectly out of phase (one compressing while the other is expanding), they cancel each other out.
Tiny microphones on the headphones pick up ambient noise. An adaptive algorithm analyzes that noise in real time, then generates an inverted copy of the waveform. This inverted signal gets played through the headphone speakers alongside your music. Where the noise wave pushes air forward, the anti-noise wave pulls it back, and the two effectively erase each other. The result is a significant reduction in perceived background noise, especially for steady, low-frequency sounds like airplane engines or air conditioning hum.
Speech Enhancement in Hearing Aids
Modern hearing aids rely heavily on digital audio processing to do more than simply amplify sound. One key technique, spectral subtraction, works by analyzing quiet moments when only background noise is present and building a profile of that noise. The processor then subtracts that noise profile from incoming audio in real time, making speech clearer against noisy backgrounds. This approach is particularly effective at pulling out the tonal components of speech from random ambient noise like crowd chatter or traffic. Spectrum shaping uses similar operations to further refine which frequencies get amplified and which get suppressed, tailoring the output to what each individual user needs to hear more clearly.
Spatial Audio and Surround Sound
Traditional surround sound is channel-based: each audio track is designed to come out of a specific speaker. A 5.1 system has six channels, each feeding one of six speakers in fixed positions. The audio is mixed with those exact speaker locations in mind.
Object-based audio, the technology behind formats like Dolby Atmos, takes a fundamentally different approach. Instead of assigning sounds to speakers, each sound is treated as an independent object with metadata describing where it should appear in three-dimensional space, along with properties like its size and how diffuse it should sound. A rendering engine then figures out how to reproduce that positioning using whatever speaker setup you actually have, whether that’s a full theater system or a pair of earbuds. This means the same audio can adapt to wildly different playback environments without being remixed.
Latency: The Speed Constraint
All digital audio processing introduces some delay, called latency, because the system needs time to capture, process, and output the signal. For most listening situations, this delay is trivially small and completely unnoticeable. But for live performance and real-time monitoring, where a singer is hearing their own voice through headphones while recording, latency becomes critical.
Professional audio systems target 10 milliseconds or less of total latency. Research has found that listeners begin noticing delay at around 15 milliseconds, particularly when they’re hearing their own voice played back. Below 10 milliseconds, the delay is generally imperceptible even to the person speaking or singing. This constraint shapes how audio hardware and software are designed: faster processors, optimized code, and smaller processing buffers all help keep latency below that perceptible threshold.