Bit Depth and Sample Rate

There is a version of this conversation that turns into a religious argument. Audiophiles insisting that 96kHz recordings have a warmth that 44.1kHz cannot touch, engineers rolling their eyes, and everyone talking past each other while the music sits forgotten. The numbers are real. What they mean is more interesting than the debate suggests.

Sample rate and bit depth measure two completely different things. Sample rate — 44.1kHz, 48kHz, 96kHz — describes how many times per second the audio is measured. 44,100 snapshots per second, or 96,000. Bit depth — 16-bit, 24-bit — describes the precision of each snapshot. Not frequency, not brightness, but dynamic range: the distance between the quietest thing the recording can capture and the loudest before it clips.

What Bit Depth Actually Does

Every bit of depth adds roughly 6dB of dynamic range. Sixteen bits gives you around 96dB. Twenty-four bits gives you around 144dB. To put that in perspective, the dynamic range of a live orchestra in a concert hall is approximately 80dB. Human hearing, in ideal conditions, tops out around 120dB before pain becomes a more pressing concern than music.

So why record at 24-bit? Not because you need 144dB of dynamic range in the final product — you don’t — but because of what happens during recording and mixing. When you’re recording a vocalist at a cautious level to avoid clipping, the signal sits in the lower registers of the dynamic range. At 16-bit, the noise floor starts to become audible when you push that signal up in the mix. At 24-bit, the noise floor is so far below what you’re working with that it’s essentially irrelevant. You have headroom to make mistakes, to push and pull levels, to process heavily without artefacts creeping in.

It’s a working format, not a listening format. The extra precision is a buffer against the thousand small decisions that accumulate across a mix.

What Sample Rate Actually Does

The Nyquist theorem says you need to sample at twice the highest frequency you want to capture. Human hearing tops out somewhere around 20kHz — for most adults, considerably lower. So 44.1kHz should be sufficient to capture everything audible, with a small margin for the anti-aliasing filters that prevent high-frequency artefacts.

This is why 44.1kHz became the standard for the CD, and why it remains the standard for most streaming distribution. It is, by any reasonable measure, enough. The argument that higher sample rates capture frequencies above human hearing, and that this somehow improves the perceived sound, has been tested repeatedly. Controlled double-blind listening tests have not been kind to the argument.

Where higher sample rates do have legitimate value is during recording and processing. At 96kHz, some analogue-modelling plugins behave more accurately. Certain saturation and harmonic distortion effects sound more natural because the artefacts they generate fall above the audible range rather than folding back into it — a phenomenon called aliasing. The difference is subtle, and whether it justifies the significantly larger file sizes and CPU load depends entirely on your setup.

48kHz sits in the middle for a practical reason: it’s the broadcast standard. Film, television, video games — anything destined for a screen traditionally runs at 48kHz. If you’re making music that will live in video, working at 48kHz from the start avoids sample rate conversion, which, while generally transparent, is one less thing to introduce any potential error.

The Working Standard

For almost any recording situation, 24-bit at 48kHz is the sweet spot. You have more than enough dynamic precision to work with freely, a sample rate that captures the full audible range with room to spare, and compatibility with both music and video pipelines. Your files are manageable. Your CPU is not screaming. You can focus on the sound instead of the settings.

When it comes time to release — whether to Spotify, Apple Music, Bandcamp, or anywhere else — 16-bit at 44.1kHz is the standard, and it is fine. The dithering process that converts your 24-bit master to 16-bit is well understood and, done correctly, transparent. The data that gets removed in that conversion was never doing anything you could hear.

The case for delivering 24-bit files to streaming services exists, and some platforms accept them, but the playback infrastructure for most listeners — earbuds on a phone, a Bluetooth speaker in the kitchen — makes the distinction academic. The difference, if it exists at all in the final listening environment, is not where most recordings need the most attention.

The numbers matter enough to understand and not so much that they should occupy the centre of your thinking. A recording made carefully at 24/48 and mixed with good judgment will almost always outperform a technically superior format applied to a careless performance. The specifications set the ceiling. Everything below it is craft.

Resonillator’s recorder and sampler modules work natively at 48kHz — open them in your browser at resonillator.com and start capturing ideas without leaving the page.