Sample Rates
44.1 vs 48 vs 96 kHz — what sample rate means, which to use, and why it matters less than you think.
Sample rate is one of those topics that generates far more debate than it deserves. The fundamentals are straightforward, the maths is settled, and the practical guidance hasn't changed in decades. Yet people still argue about it.
Here's what actually matters.
Specs
What Sample Rate Means
A sample rate is how many times per second a digital audio system captures the analogue signal. At 44.1 kHz, the converter takes 44,100 snapshots of the waveform every second. At 48 kHz, it takes 48,000.
The Nyquist–Shannon sampling theorem states that to accurately capture a frequency, you need to sample at least twice that frequency. So 44.1 kHz captures frequencies up to 22.05 kHz. 48 kHz captures up to 24 kHz. Human hearing tops out at roughly 20 kHz — and for most adults, significantly less.
This isn't a simplification. It's the maths. A correctly sampled and reconstructed signal is a perfect representation of the original within its frequency band. Not "close enough" — mathematically perfect. The common misconception that digital audio is made of "staircase steps" is wrong; the reconstruction filter smooths the output into a continuous waveform.
Why 44.1 kHz Exists
44.1 kHz was chosen as the CD standard (Red Book, 1980) because it comfortably covers the full range of human hearing with a small margin above 20 kHz for the anti-aliasing filter to roll off. The specific number has roots in early digital recording equipment that used video tape as a storage medium — the maths worked out neatly with PAL and NTSC frame rates.
48 kHz became the standard for video and film production. If you're working with any visual media, 48 kHz is non-negotiable. Broadcast, film, television, YouTube — all 48 kHz.
The two standards exist for historical reasons, not because one is meaningfully better than the other for audible quality.
Higher Sample Rates
88.2, 96, and 192 kHz exist, and they have their advocates. The arguments in favour:
- Processing headroom. Some plugin algorithms — particularly EQs, compressors, and saturation — behave more accurately at higher sample rates because there's more frequency space above the audible band for filter curves to settle into. This is a real technical benefit, though many modern plugins oversample internally to achieve the same result.
- Better anti-aliasing filter behaviour. A wider gap between the top of the audible band and the Nyquist frequency means the reconstruction filter can have a gentler slope, which some argue produces less phase distortion near 20 kHz.
- Ultrasonic harmonics. Some engineers believe that capturing harmonics above 20 kHz — from cymbals, strings, brass — contributes to a sense of "air" or realism, even if those frequencies aren't directly audible. The evidence for this is contested.
The arguments against:
- It's inaudible. Controlled blind tests consistently fail to demonstrate that listeners can distinguish 44.1 kHz from 96 kHz recordings of the same source.
- File sizes double or quadruple. A 96 kHz session uses twice the disk space and processing power of a 48 kHz session. At 192 kHz, four times.
- Diminishing returns. The theoretical benefits are marginal at best and irrelevant for the vast majority of recording and mixing work.
The honest answer: for most work, 44.1 or 48 kHz is perfectly adequate. Higher sample rates have some genuine technical benefits at the processing stage, but they won't make your recordings sound noticeably better to anyone listening on headphones or speakers.
Practical Guidance
- Recording: 48 kHz / 24-bit is the sensible default. It covers all delivery formats, works with video, and gives you more than enough frequency bandwidth.
- Working with video: 48 kHz. Always. No exceptions.
- Mastering for CD: 44.1 kHz final delivery (Red Book specification).
- Mastering for streaming: 44.1 or 48 kHz. Spotify, Apple Music, and YouTube all accept both. There is no audible advantage to delivering at a higher rate — the services will transcode to lossy formats anyway.
- Don't record at 192 kHz unless you have a specific, articulated reason and the storage to support it. "Just in case" is not a reason. Your interface's converters may also perform worse at their maximum sample rate than at 48 or 96 kHz.
Sample Rate Conversion
Changing sample rates after recording — sample rate conversion (SRC) — is a necessary evil. Converting from 48 kHz to 44.1 kHz for CD delivery, for example, requires resampling the entire waveform. Done well, it's transparent. Done poorly, it introduces subtle artefacts: aliasing, ringing, or loss of high-frequency detail.
Use high-quality SRC algorithms. SoX (free, open-source) is excellent. iZotope's SRC is widely trusted in mastering. Most DAWs have built-in SRC that ranges from acceptable to very good — check your DAW's documentation for its quality settings.
The best approach: record at your target delivery rate whenever possible. If you're delivering at 44.1 kHz, record at 44.1 kHz (or 88.2 kHz, which divides down cleanly). If you're working with video, record at 48 kHz and stay there.
Relationship to Bit Depth
Sample rate and bit depth are often mentioned together, but they control entirely different things:
- Sample rate determines the highest frequency that can be captured. It controls frequency resolution.
- Bit depth determines the dynamic range — the gap between the quietest and loudest signals the system can represent. 16-bit gives you 96 dB of dynamic range. 24-bit gives you 144 dB.
In practice, bit depth makes a bigger difference than sample rate for most recording work. The jump from 16-bit to 24-bit gives you vastly more headroom to record at conservative levels without burying your signal in the noise floor. The jump from 48 kHz to 96 kHz gives you frequency content above 24 kHz that almost nobody can hear.
Record at 24-bit. Always. The storage cost is negligible compared to the benefit. Whether you record at 44.1, 48, or 96 kHz matters far less.