Sample rate and bit depth explained: what to record at, and what to deliver

Every audio file carries two numbers that producers argue about endlessly: the sample rate (44.1 kHz, 48 kHz, 96 kHz) and the bit depth (16-bit, 24-bit, 32-bit float). Forums will tell you that higher is always better, that 96 kHz is "pro", and that your distributor will reject anything less. Most of that is folklore. This guide sets out what the two numbers do, what the two biggest platforms actually ask you to deliver, and what we would do on a real independent release.

Everything below is as we read the platforms' own pages on 24 Oct 2026. Specs change, so check the linked page before a big release. If you are already wondering about loudness, our guide to what LUFS to master to covers the other half of delivery.

What the two numbers mean

Digital audio is a series of snapshots of a waveform. Apple's Apple Digital Masters guide puts it plainly: the resolution of a recording is set by "the sample rate (how many times per second samples are taken) and the bit depth (how many bits are used to represent each sample)".

• Sample rate decides the highest frequency a file can hold. The same Apple guide explains the Nyquist rule: to represent a signal you need a sampling rate double its highest frequency. Human hearing tops out around 20 kHz, so you need more than 40 kHz, and Apple says the CD's 44.1 kHz "is adequate for this need".
• Bit depth decides how finely each snapshot is measured, which in practice means the noise floor and how much dynamic range you have. Apple's guide notes that CD audio at 16-bit gives each sample a value between -32,768 and 32,767, and that 24-bit recording "has become the norm", increasing "workable headroom and total dynamic range".

Two things are often confused with these. Bit rate is a different measure: Apple's guide is explicit that bit rate is not bit depth, and it is what a compressed stream like AAC is described in. And 32-bit float is a working format inside your DAW, not usually a delivery format; our piece on gain staging in 32-bit float explains what it does and does not protect you from.

What Spotify asks for

Spotify's artist help page on audio file formats is short and specific. As we read it:

Note what is missing: there is no instruction to deliver at 96 kHz or any specific high rate. The floor is 44.1 kHz, and the bit depth rule is about honesty, not size. If your master was made at 24-bit, send 24-bit. Most artists deliver through a distributor rather than to Spotify directly, so your distributor's upload form may add its own rules on top; follow theirs where they are stricter. Our earlier piece on what to deliver now Spotify streams lossless covers the distributor side.

What Apple asks for

Apple's guidance is the most detailed of any major platform, and it lives in the Apple Digital Masters document rather than a short help page. The key lines, quoted from the guide:

• "use only 24-bit sources and send us the highest-resolution master file possible, appropriate to the medium and the project."
• "Don't upsample files to a higher resolution than their original format. Upsampling won't recover or add information to an audio file."
• Don't "bit-pad" or recapture 16-bit files in 24-bit.
• Some engineers prefer to do their own sample rate conversion, "however we ask that you deliver the highest native sample rate available."

Apple's guide also explains why it wants the high-resolution file. When the source rate is higher than the production rate, Apple says it downsamples with its own "mastering-quality" sample rate conversion into a 32-bit floating-point file, and that this lets its encoders use the full dynamic range of a 24-bit source and eliminates "the need for adding dither". In other words, Apple would rather do the final reduction itself than receive a file someone has already squeezed down to 16-bit.

On the listening side, Apple's support page about lossless audio is a useful reality check: it says the 14-inch and 16-inch MacBook Pro support native playback "up to 96 kHz", and that on other Macs, iPhone and iPad you need an external digital-to-analogue converter to play songs at sample rates higher than 48 kHz. Most of your listeners will not hear your 96 kHz file at 96 kHz.

Dither, downsampling and the mistakes that cost you

Two processes sit between your session and a delivery file, and Apple's guide defines both.

• Downsampling is using sample rate conversion to lower the rate, for example from 96 kHz to 48 kHz.
• Dither is added when you reduce bit depth, for example from 24-bit to 16-bit. Apple describes it as "a tradeoff": distortion is reduced at the cost of added noise. The alternative, truncating, simply drops the extra bits.

From the two platforms' rules, the practical mistakes are easy to list:

1. Upsampling a 44.1 kHz session to 96 kHz before delivery. Apple says it adds nothing, and Spotify only asks for 44.1 kHz or higher. You gain a bigger file and no new information.
2. Padding a 16-bit file to 24-bit. Apple names this explicitly. If the only master you have is 16-bit, Spotify's page says to send 16-bit.
3. Dithering down to 16-bit when 24-bit is accepted. Both platforms accept 24-bit, and Apple says its pipeline avoids dither on a 24-bit source. Keep a 16-bit, dithered file for CD or for any outlet that specifically asks for one.

A simple plan for a real release

None of the platforms above tells you what to record at, so this part is our recommendation, built on what they ask you to deliver.

• Record and mix at 24-bit (or 32-bit float inside the DAW). This matches what both Spotify and Apple prefer for the master.
• Pick one sample rate at the start and keep it. 44.1 kHz or 48 kHz is fine for music; both meet Spotify's floor. Choose a higher rate if your plug-ins, your converters or a sync or video client genuinely need it, not because a forum said so.
• Deliver at the session's native rate and 24-bit, as FLAC or WAV, stereo. Do not convert up, and do not convert down unless your distributor's form requires it.
• Archive the native master. Apple's argument for keeping the highest-resolution master is that it keeps future formats open. That applies to your own hard drive too.

What we think

The sample rate debate is mostly a status game, and it costs independent artists time and storage they do not have. The two platform specs we read are notably relaxed: Spotify sets a 44.1 kHz floor, and Apple's strictest rules are about not faking resolution, not about chasing it. Neither says a 96 kHz master will be treated better.

So our advice is boring on purpose. Spend your effort on the arrangement, the mix and the loudness decisions, which listeners actually hear, and keep the file format honest: 24-bit, your native rate, no upsampling, no padding. We would also like distributors to show the same clarity Spotify does, in one short table on the upload page, so artists stop guessing and stop bouncing five versions of the same master "just in case". The rule that protects artists best is the one Apple wrote down: deliver what you actually made.