Sampling rate is the number of times per second a device measures a continuous signal, like sound or a heartbeat, to convert it into digital data. It’s expressed in hertz (Hz), where 1 Hz equals one measurement per second. A higher sampling rate captures more detail, while a lower one saves storage space and energy. Whether you’re recording music, monitoring your heart rate, or just wondering why your audio files are a certain size, sampling rate is the concept that ties it all together.
How Analog Signals Become Digital
Sound, light, and electrical signals from your body are all continuous. They flow smoothly through time without breaks. But computers can’t process a smooth, infinite stream of information. They need individual data points, discrete numbers they can store and manipulate.
Sampling is the process of slicing that continuous signal into tiny snapshots taken at regular intervals. Each snapshot captures the signal’s value at one specific instant. String enough of these snapshots together and you get a convincing digital replica of the original. Think of it like a flipbook: each page is a still image, but flip through them fast enough and you see smooth motion. The sampling rate determines how many “pages” you get per second. More pages, smoother motion.
The Rule That Makes It All Work
There’s a hard mathematical requirement for how fast you need to sample. Known as the Nyquist theorem, it states that your sampling rate must be at least twice the highest frequency you want to capture. If you’re recording audio and want to preserve frequencies up to 20,000 Hz (the upper limit of human hearing), you need a sampling rate of at least 40,000 samples per second.
This isn’t a suggestion. It’s a threshold. Drop below it and the digital version of your signal won’t just lose detail, it will introduce entirely false information called aliasing.
What Happens When the Rate Is Too Low
Aliasing is the most practically important consequence of an insufficient sampling rate, and it’s stranger than simple quality loss. When a frequency in the original signal is higher than half the sampling rate (the Nyquist frequency), that frequency doesn’t just disappear. It “folds back” and shows up as a completely different, lower frequency that wasn’t in the original signal at all.
Here’s a concrete example. If you sample at 8,000 Hz, your Nyquist frequency is 4,000 Hz. A pure 3,000 Hz tone will be captured accurately because it falls below that limit. But if a 5,000 Hz tone comes in, the system records it as a 3,000 Hz tone instead, because the alias appears at the sampling rate minus the input frequency (8,000 minus 5,000 equals 3,000). You’d play it back and hear entirely the wrong pitch with no way to tell it’s wrong from the data alone.
With complex sounds, the problem gets worse. A note at 1,500 Hz with harmonics at 3,000, 4,500, and 6,000 Hz has a clean frequency ratio of 1:2:3:4. Sampled at 8,000 Hz, the 4,500 Hz harmonic folds to 3,500 Hz and the 6,000 Hz harmonic folds to 2,000 Hz. The result contains frequencies at 1,500, 2,000, 3,000, and 3,500 Hz, a ratio that doesn’t match the harmonic series at all. The sound takes on an unnatural, metallic, or dissonant quality that wasn’t present in the original.
Why CDs Use 44,100 Hz
The CD sampling rate of 44,100 Hz seems oddly specific, and its origin is one of audio’s best quirks. Human hearing tops out around 20,000 Hz, so the Nyquist theorem demands a rate above 40,000 Hz. The extra headroom accounts for the fact that real-world filters used to prevent aliasing aren’t perfect and need some margin to work effectively. But why 44,100 exactly?
In the early days of digital audio, storing roughly 1 megabit per second per channel was a serious challenge. Hard drives had the speed but not the capacity for long recordings, so engineers repurposed video recorders. They encoded audio samples as black and white levels in a fake video signal, storing data on each line of a video frame. The sampling rate had to divide evenly into the structure of television standards. Both the American 525-line/60 Hz system and the European 625-line/50 Hz system could produce exactly 44,100 samples per second with the right line and field parameters. The math worked out cleanly for both: 60 times 245 active lines times 3 samples per line equals 44,100, and 50 times 294 active lines times 3 samples equals the same number. When CDs arrived, they inherited this rate from the mastering equipment, even though CDs themselves contain no video circuitry.
Digital audio tape (DAT) later adopted 48,000 Hz, which divides more neatly into the 8,000 and 16,000 Hz rates used in telephone audio. One persistent claim is that the two rates were kept deliberately incompatible to make direct digital copying between DAT and CD more difficult.
Sampling Rate vs. Bit Depth
These two specs often appear side by side on audio equipment, and people frequently confuse them. They control completely different things. Sampling rate determines the highest frequency you can capture. That’s it. A higher sampling rate doesn’t make the audio louder, cleaner, or more dynamic on its own. It simply extends the frequency ceiling.
Bit depth, on the other hand, controls how precisely each individual sample measures the signal’s volume at that instant. Higher bit depth means more possible volume levels for each snapshot, which pushes the noise floor lower. A 16-bit recording (CD standard) has a different noise floor than a 24-bit recording (professional standard), but both can capture the same range of frequencies if their sampling rates match.
Think of sampling rate as how often you take a photo and bit depth as the resolution of each photo. More photos per second capture faster motion. Higher resolution per photo captures finer detail in each frame. You need both to get a high-quality result, but they solve different problems.
Sampling Rates in Health Monitoring
The same principles apply to wearable sensors and medical devices, just with different numbers. Clinical electrocardiograms (ECGs) that track the heart’s electrical activity typically sample at 1,000 Hz, capturing a thousand measurements per second. That high rate preserves the sharp, fast electrical spikes that cardiologists need to see when analyzing heart rhythm.
Consumer wrist-worn heart rate monitors use optical sensors (called PPG sensors) that work by shining light through your skin and measuring blood flow changes. These devices generally sample between 21 and 64 Hz. The Empatica E4, a commonly studied research wristband, records at 64 Hz. That range is enough for tracking heart rate and basic heart rate variability, but it’s orders of magnitude lower than clinical ECG because the optical signal is smoother and the diagnostic needs are simpler.
Choosing a sampling rate in wearable devices is always a trade-off. Higher rates produce more precise data but drain the battery faster and fill up storage sooner. For a device you’re wearing all day, 64 Hz is often the practical ceiling.
How Sampling Rate Affects File Size
The relationship is straightforward: double the sampling rate and you double the file size, all else being equal. Uncompressed audio file size is calculated by multiplying the bit depth (divided by 8 to convert bits to bytes) by the sampling rate by the duration in seconds. For stereo, you double the result since there are two channels.
A one-minute stereo recording at CD quality (44,100 Hz, 16-bit) produces about 10.6 megabytes of raw data. Bump that to 96,000 Hz at 24-bit, a common high-resolution format, and the same minute balloons to roughly 34.6 megabytes. Compression formats like MP3 and AAC reduce these sizes dramatically, but the raw data cost of a higher sampling rate is always linear. If you’re recording hours of audio or streaming data from dozens of sensors, that multiplication adds up quickly.