Choosing Sample Rate and Bit Depth: 44.1 vs 48 vs 96 kHz

The Two Numbers That Define Your Audio Quality

Every digital audio file has two fundamental technical parameters: sample rate and bit depth. These numbers determine the frequency range your recording can capture and the dynamic range it can preserve. Understanding what they actually do, rather than following marketing claims, lets you make informed decisions at the very start of every project.

This guide is part of our Recording and Session Prep series. Getting these settings right from the beginning means you never have to worry about sample rate conversion or bit depth truncation partway through a project. Choose once, choose correctly, and focus on making music.

What Sample Rate Determines

Sample rate is the number of times per second that the analog-to-digital converter measures (samples) the incoming audio signal. A sample rate of 44,100 Hz means the converter captures 44,100 snapshots of the audio waveform every second. A sample rate of 96,000 Hz captures 96,000 snapshots per second.

The Nyquist Theorem Simplified

The Nyquist-Shannon sampling theorem states that a digital system can accurately capture any frequency up to half its sample rate. This maximum capturable frequency is called the Nyquist frequency. At 44.1 kHz, the Nyquist frequency is 22,050 Hz. At 48 kHz, it is 24,000 Hz. At 96 kHz, it is 48,000 Hz.

Human hearing ranges from approximately 20 Hz to 20,000 Hz, with most adults losing sensitivity above 16,000 Hz by their mid-twenties. A sample rate of 44.1 kHz captures frequencies up to 22,050 Hz, which covers the entire audible spectrum with a small safety margin. This is why 44.1 kHz was chosen as the standard for CD audio: it captures everything a human can hear.

Common Sample Rates and Their Uses

• 44.1 kHz: The standard for music production. CDs, streaming platforms (Spotify, Apple Music, Tidal standard), and most digital distribution services use 44.1 kHz. If your final destination is music streaming or CD, recording at 44.1 kHz avoids any sample rate conversion.
• 48 kHz: The standard for video and broadcast audio. If your music is intended for film, television, YouTube, or any video platform, 48 kHz is the correct choice. Video editing software and broadcast standards are built around 48 kHz.
• 88.2 kHz and 96 kHz: Sometimes used for recording and mixing sessions where the producer wants extra frequency headroom for plugin processing. Some engineers argue that certain plugins (especially saturation, EQ, and dynamics) produce fewer aliasing artifacts at higher sample rates. However, most modern plugins internally oversample to handle this, making the external sample rate less relevant.
• 192 kHz: Rarely necessary for music production. Used in specialized scientific and archival applications. The file sizes are enormous (roughly four times larger than 48 kHz), CPU load on every plugin quadruples, and there is no audible benefit for music listeners.

What Bit Depth Determines

Bit depth determines the number of possible amplitude values available to represent each sample. More bits means more values, which means finer resolution in capturing the volume of the audio at each snapshot. The practical result is a larger usable dynamic range, which is the difference between the quietest sound you can capture and the loudest sound before clipping.

Bit Depth and Dynamic Range

• 16-bit: 65,536 possible amplitude values per sample. Theoretical dynamic range: 96 dB. This is the standard for CD delivery and is adequate for the final listener, but leaves limited headroom during recording and mixing.
• 24-bit: 16,777,216 possible amplitude values per sample. Theoretical dynamic range: 144 dB. This massive headroom means you can record at conservative levels without approaching the noise floor, giving the mix engineer maximum flexibility.
• 32-bit float: Used internally by most DAW processing engines. Provides effectively unlimited dynamic range within the DAW, meaning internal clipping is nearly impossible. However, 32-bit float is a processing format, not a recording format; your audio interface still converts at 24-bit (or 16-bit).

Why 24-Bit Is Always the Right Choice for Recording

Record at 24-bit on every project without exception. The additional dynamic range costs you only 50 percent more file size compared to 16-bit (a 3-minute stereo WAV at 44.1 kHz is approximately 30 MB at 16-bit and 45 MB at 24-bit). That extra 15 MB per file gives you 48 dB of additional dynamic range, which translates to dramatically more headroom during recording and mixing.

16-bit should only appear at the very end of the production chain: when you export the final master for CD delivery. All recording, mixing, and mastering should happen at 24-bit or higher.

When Higher Sample Rates Actually Matter

The argument for higher sample rates in production is not about capturing ultrasonic frequencies that humans cannot hear. It is about how digital processing algorithms behave near the Nyquist frequency.

Some plugins, particularly those that model nonlinear analog processes like saturation, tape emulation, and tube compression, generate harmonic content above the original signal frequencies. At 44.1 kHz, harmonics that exceed 22,050 Hz are "folded back" into the audible range as aliasing artifacts, which can sound harsh or metallic. At 96 kHz, the Nyquist frequency is 48,000 Hz, giving those harmonics more room before aliasing occurs.

However, most professional plugins released in the last decade include internal oversampling that handles this problem automatically. They temporarily upsample the audio within the plugin, process it, and downsample the result, all transparent to you. This means the external session sample rate matters less than it did ten years ago, and 44.1 kHz or 48 kHz is perfectly clean for modern production.

For a deeper understanding of how recording levels interact with these digital characteristics, our session cleanup guide covers how to organize and prepare your project files once you have committed to your technical settings.

Dithering When Converting Bit Depth

Dithering is a process that adds a very small amount of shaped noise when reducing bit depth, typically when converting from 24-bit to 16-bit for final delivery. Without dithering, reducing bit depth truncates the lowest bits, creating a form of distortion called quantization noise. This noise is correlated with the audio signal and sounds unnatural, particularly on quiet passages and fade-outs.

Dithering replaces this correlated distortion with uncorrelated random noise that sounds like a very faint, smooth hiss. The dithered noise is perceptually less objectionable than quantization distortion, and at normal listening levels, it is completely inaudible.

When to Apply Dithering

• Apply dithering when bouncing the final master from 24-bit to 16-bit for CD or specific distribution requirements.
• Do not apply dithering when exporting stems at 24-bit, when bouncing between 24-bit sessions, or when delivering a 24-bit master file to a streaming platform (most platforms accept 24-bit).
• Apply dithering only once, at the very last stage of production. Multiple rounds of dithering accumulate noise.

Common dither algorithms include POW-r (types 1, 2, and 3), MBIT+, and flat triangular (TPDF). For music, POW-r type 2 or 3 is a popular choice because it shapes the dither noise into frequency ranges where human hearing is least sensitive, making the noise virtually imperceptible.

File Size Implications

Higher sample rates and bit depths produce larger files. Here are approximate sizes for one minute of stereo audio:

• 44.1 kHz, 16-bit: approximately 10.1 MB per minute
• 44.1 kHz, 24-bit: approximately 15.1 MB per minute
• 48 kHz, 24-bit: approximately 16.5 MB per minute
• 96 kHz, 24-bit: approximately 33.0 MB per minute
• 192 kHz, 24-bit: approximately 66.0 MB per minute

For a 30-track session of a 4-minute song at 44.1 kHz and 24-bit, you are looking at roughly 1.8 GB of raw audio. At 96 kHz and 24-bit, that same session balloons to approximately 3.96 GB. Factor in backup copies and archive storage, and the difference compounds significantly over a catalog of projects.

Storage is cheap, but not free. CPU load is the more practical concern: doubling the sample rate roughly doubles the processing load on every plugin in the session. A session that runs smoothly at 44.1 kHz may stutter and drop audio at 96 kHz if your computer is not powerful enough to handle the increased throughput.

The "Higher Is Always Better" Myth

Marketing from audio equipment manufacturers often implies that higher sample rates and bit depths always produce better results. This is misleading. Beyond the audible threshold, higher numbers primarily increase file sizes, CPU load, and storage costs without delivering perceptible improvements to listeners.

Controlled blind listening tests consistently show that trained audio engineers cannot reliably distinguish between 44.1 kHz and 96 kHz recordings of the same source. The Nyquist theorem is not a suggestion; it is a mathematical proof. A properly implemented 44.1 kHz recording captures the full audible frequency range without loss.

Where bit depth does matter, 24-bit is decisively better than 16-bit for recording and mixing. But 32-bit recording (as opposed to 32-bit float internal processing) is unnecessary because no commercially available analog-to-digital converter actually achieves more than 21 to 22 effective bits of dynamic range anyway. The extra bits in a 24-bit file are already below the converter's noise floor.

Make your decisions based on your delivery target (music streaming, video, archival) and your system's capabilities. For setting up the right recording levels within these chosen settings, our gain staging guide shows you exactly where your levels should land at every point in the chain.