Sample Rate and Bit Depth: What to Use at Each Stage

Sample rate is how many times per second the audio is measured. Bit depth is how finely each measurement is stored. Record and mix at 24-bit, at 44.1 kHz or 48 kHz, and deliver the resolution you have rather than converting down first. Apple Digital Masters asks for 24-bit sources, and YouTube's Content ID partner specification recommends 24-bit for music-video deliveries.

Sample rate and bit depth are the two numbers your DAW asks for before you record a note, and most producers take the default and move on. That works most of the time. It stops working at the edges: a plugin adding harmonics nobody asked for, a quiet passage with more hiss under it than it should have, a delivery spec that turns out to want something other than what you bounced.

This guide covers what each number does mechanically, what the published listening tests found when they went looking for an audible difference, and what the major platforms currently ask you to hand over. For where these settings sit in the wider workflow, see Music Production Basics.

Sample Rate: How Often the Wave Gets Measured

Digital audio stores a sound wave as a long list of amplitude measurements. Sample rate is how many of those measurements happen per second, written in kilohertz. At 44.1 kHz, the converter measures the incoming signal 44,100 times every second.

The sampling theorem sets the floor. To reconstruct a frequency you need a sample rate greater than twice that frequency, so a 44.1 kHz file can carry frequencies up to 22.05 kHz. Textbook human hearing runs from roughly 20 Hz to 20 kHz, and the top of that range narrows with age. That is where 44.1 kHz comes from: it clears 20 kHz with a little room left for the converter's filter.

The Red Book CD standard fixed 44.1 kHz at 16-bit as the delivery pair, which is why those two numbers still show up as defaults in places that have nothing to do with CDs.

What Each Rate Costs You

Storage is one cost. Higher session rates also tend to raise CPU load because the DAW and many plug-ins process more samples per second, although some plug-ins use their own internal oversampling.

Does a Higher Sample Rate Sound Better?

Most guides pick a side here. The published tests do not agree with each other, and it is worth knowing what each found before you spend disk space on the argument.

Meyer and Moran ran double-blind tests over a year, inserting a 16-bit/44.1 kHz stage into high-resolution playback (Journal of the Audio Engineering Society, September 2007). Their reported result: the CD-quality loop was undetectable at normal-to-loud listening levels, across their subjects and their systems.

Reiss later pooled 18 published experiments covering more than 400 participants and over 12,500 trials (Journal of the Audio Engineering Society, June 2016). That meta-analysis reported a small but statistically significant ability to discriminate high-resolution material, and said the effect increased dramatically when subjects received extensive training.

Both are real results pointing in different directions. The careful reading is that any difference at playback is small and hard to hear without training, and that the question is still open. This page is not going to close it for you.

Where a Higher Rate Does Something Measurable

Two mechanisms are worth knowing, and both are about the machinery rather than the ear.

The anti-alias filter. An anti-aliasing filter band-limits the input before conversion. In a traditional Nyquist-rate converter, higher rates widen the transition band; many modern audio converters instead oversample internally and combine analog with digital filtering, so the output rate alone does not determine the analog filter's steepness.

Aliasing inside plugins. Saturation, distortion, clipping and other nonlinear processes generate harmonics above the original signal. Any harmonic that lands above the Nyquist frequency folds back down into the audible range as tones that were never in the source. Running the session at a higher rate pushes those products further out of the way. Many modern plugins oversample internally for the same reason, which gets you the benefit without doubling the CPU and storage load of the whole session.

Practical starting point: run sessions at 48 kHz. It matches video and broadcast work and matches YouTube's recommendation for Content ID partner music-video deliveries. If your machine is struggling, 44.1 kHz is a fine place to work.

Bit Depth: How Finely Each Measurement Is Stored

Bit depth is the resolution of each individual sample: how many values the converter can choose from when it writes down an amplitude. 16-bit gives 65,536 values. 24-bit gives 16,777,216.

That resolution becomes dynamic range, the distance between the loudest signal the format can hold and the noise floor underneath it, at roughly 6 dB per bit.

Why Bit Depth Is the One to Get Right

Recording at 24-bit buys you permission to leave headroom. Set peaks around -12 dBFS and there is still a large margin above the noise floor, so nothing has to be pushed toward 0 dBFS to stay clean. That is the same argument as gain staging, approached from the other end.

One caveat worth stating plainly. The 144 dB figure is theoretical. Real recordings meet the noise floor of the room, the mic and the preamp long before they meet the noise floor of a 24-bit file. 24-bit takes the format out of the equation; it does not make a noisy room quiet.

32-bit float is a different thing again. It is the internal format most DAWs mix in, which is why a channel that goes past 0 dBFS inside the box can usually be pulled back down without damage. Some interfaces and field recorders now capture in 32-bit float as well, which lowers the price of getting input gain wrong at the source.

Dither: Only When the Bit Depth Drops

Truncating 24-bit audio to 16-bit discards the bottom 8 bits. Done bluntly, the rounding errors correlate with the signal, which shows up as a low-level distortion that tracks the music instead of a steady hiss. Dither adds a tiny amount of noise before the rounding, decorrelating the error and turning that distortion into a constant, harmless noise floor.

The short version:

• Consider dither whenever the rendered signal is reduced to a lower fixed-point word length. A 16-bit delivery is the common case, and some DAWs also recommend dither for a 24-bit render from higher-precision internal processing.

• Do not infer the dither setting from the session-file bit depth alone; check the DAW's internal processing depth and export guidance.

• For stems, preserve enough resolution for the next stage. Some DAWs recommend 32-bit float to avoid dithering before further processing; when the receiving workflow requires 24-bit, follow the DAW's export guidance.

• Apply it once, at the final step. Stacked dither stages stack their noise.

Noise-shaped options push the dither noise up toward the frequencies where hearing is least sensitive. Logic's dither menu, for example, lists POWr #1 as plain dithering and POWr #2 and #3 as noise-shaping variants. Which one suits a given master is a judgment call about the material, not a setting with one right answer.

What the Platforms Ask You to Deliver

This is where the old studio rule has drifted away from the published specs. "Bounce everything to 16-bit/44.1 kHz" made sense when CD was the destination. The current delivery documents read differently. Checked August 2026 against each platform's own material:

The practical version: keep a 24-bit master at the session's native sample rate as your delivery source, and make a 16-bit/44.1 kHz version when something specifically asks for one. Mastering for Streaming covers the loudness half of the same delivery question, and the Music Distribution Guide covers who you hand the file to.

Settings by Stage

If the choice is between a higher sample rate and a better room, mic or mix, Orphiq's guides for artists keep landing on the same answer. The room and the performance change the result more than the number in the session dialog does.

Frequently Asked Questions

Should I record at 44.1 kHz or 48 kHz?

Either works. 48 kHz is the more flexible default: it matches video work and YouTube's recommendation for Content ID partner music-video deliveries. If CPU or disk space is tight, 44.1 kHz costs you little.

Does 24-bit recording make the finished track sound better?

It gives you more room to work: a lower noise floor, more headroom, less pressure on input levels. Whether a listener hears that in the master depends on everything else in the chain.

Do streaming services support high-resolution audio?

Spotify offers lossless FLAC up to 24-bit/44.1 kHz for eligible Premium listeners. Apple Music offers ALAC, with Hi-Res Lossless reaching 24-bit/192 kHz. Both need the right playback setup.

Should I still bounce a 16-bit/44.1 kHz master?

Make one when a distributor, a CD plant or a sync brief asks for it. As an automatic default, it now runs counter to Apple Digital Masters guidance and YouTube's Content ID partner recommendation for music-video deliveries.

Read Next

• Music Production Basics: From Bedroom to Release

• Gain Staging Explained for Producers

• Mastering for Streaming: Loudness and Format Guide

• Audio Stems Explained: What They Are and Why They Matter

• Mixing vs Mastering: What Each Step Does

From Session to Release:

Session settings are a one-time decision. What comes out of the session still needs a date and a plan around it. How to Plan a Music Release: Step-by-Step Checklist is the next step once you have a finished master.