Learn
Sample Rate and Bit Depth Explained
Sample rate and bit depth are the two fundamental settings that define the quality of digital audio. Every digital recording, from a voice memo on your phone to a multitrack studio session, is shaped by these two numbers. They determine how faithfully a digital file captures the original sound, how much storage it requires, and what you can do with it in post-production.
Despite their importance, these concepts are often misunderstood or over-complicated. The underlying principles are simple, and the practical choices they lead to are simpler still.
What sample rate means
Sound in the physical world is a continuous wave of air pressure changes. Digital audio captures that wave by measuring its amplitude at regular intervals, each measurement called a sample. The sample rate is the number of these measurements taken per second, expressed in hertz (Hz) or kilohertz (kHz).
At a sample rate of 44,100 Hz (44.1 kHz), the audio system takes 44,100 snapshots of the sound wave every second. At 48 kHz, it takes 48,000. The higher the sample rate, the more frequently the wave is measured, and the more accurately the digital representation tracks the original continuous signal.
The reason sample rate matters comes down to a principle called the Nyquist-Shannon sampling theorem. In simplified terms, a digital system can accurately capture frequencies up to half its sample rate. A 44.1 kHz recording can represent frequencies up to approximately 22.05 kHz. A 48 kHz recording extends that ceiling to 24 kHz. Since human hearing tops out at roughly 20 kHz (and often lower, particularly with age), both of these standard rates capture everything a person can hear.
This is not a limitation of the technology; it is a mathematical certainty. The sampling theorem is not an approximation or a rule of thumb. When the sample rate is at least double the highest frequency present, the reconstruction is perfect. No information is lost.
Standard sample rates and when to use them
44.1 kHz
This is the sample rate of the compact disc, established in the early 1980s and still the standard for music distribution. It captures frequencies up to about 22 kHz, comfortably exceeding the range of human hearing. If your audio will be distributed as music, whether as downloads, CDs, or streams, 44.1 kHz is the native target format. Most streaming services accept higher sample rates but transcode down to 44.1 kHz for delivery.
48 kHz
The broadcast and video standard. Film, television, games, and online video all use 48 kHz as their default audio sample rate. If your audio will accompany video of any kind, 48 kHz is the correct choice. This avoids the need for sample rate conversion later in the workflow, which, while generally transparent, is an unnecessary step if you can avoid it.
For most content creators and video editors, 48 kHz is the sensible default for everything. It covers both music and video use cases without compromise, and the marginal increase in file size compared to 44.1 kHz is negligible.
96 kHz
Double the standard rates, 96 kHz captures frequencies up to 48 kHz, well beyond what any human ear can perceive. It is used in some professional mastering workflows and high-resolution audio releases. The practical argument for 96 kHz during recording is not about capturing ultrasonic content but about giving certain digital processing algorithms (resampling, pitch shifting, time stretching) more data to work with, potentially yielding cleaner results.
The file size doubles compared to 48 kHz at the same bit depth, so working at 96 kHz has meaningful storage and bandwidth implications when managing large sessions.
192 kHz
The highest sample rate in common use, 192 kHz quadruples the file size relative to 48 kHz. It exists, and some audiophile releases are sold at this rate, but its practical benefits for listening are essentially zero. No controlled study has reliably demonstrated that listeners can distinguish 192 kHz audio from 96 kHz or even 48 kHz in blind tests. It is occasionally used in specialised scientific or archival contexts where capturing ultrasonic information has value, but for music production and consumption, it is overkill.
What bit depth means
While sample rate determines how often the sound wave is measured, bit depth determines how precisely each measurement is recorded. Specifically, bit depth defines the number of possible amplitude values each sample can take.
Think of it this way: if you were asked to describe the height of a wave using only whole numbers from 0 to 10, you could capture the general shape but would lose fine detail. If you could use numbers from 0 to 65,535, your description would be far more precise. Bit depth works on the same principle, using powers of two.
At 16 bits, each sample can take one of 65,536 possible values. At 24 bits, that number jumps to 16,777,216. More possible values means greater precision in representing the amplitude of the audio signal, which translates directly into two practical qualities: dynamic range and noise floor.
Dynamic range is the span between the quietest and loudest sounds a system can represent. A 16-bit recording offers a theoretical dynamic range of roughly 96 dB, which is considerable; it comfortably exceeds the dynamic range of most recorded music. A 24-bit recording extends this to approximately 144 dB, which is beyond the threshold of pain and well past what any real-world recording environment can achieve.
The noise floor is the level of background noise inherent in the digital system itself (quantisation noise). Higher bit depth pushes this noise floor lower, making it less likely to interfere with quiet passages in the audio.
Standard bit depths and their roles
16-bit
The bit depth of the compact disc and the standard for finished consumer audio for decades. A 96 dB dynamic range is more than adequate for the final delivery of music, podcasts, and most audio content. Virtually all music you have ever listened to from a CD, a download store, or a streaming service was either recorded in 16-bit or dithered down to 16-bit for distribution.
For playback and distribution, 16-bit remains perfectly sufficient. The claim that 16-bit audio sounds "flat" or "harsh" compared to 24-bit reflects misunderstandings about dithering and mastering practices, not a genuine limitation of the format for finished, well-produced audio.
24-bit
The professional recording and mixing standard. The additional dynamic range that 24-bit provides is not about making the final product sound better to the listener; it is about giving the recording engineer headroom during capture and processing.
When recording live performances or acoustic instruments, signal levels are unpredictable. At 24 bit, you can set conservative input levels without worrying that quiet passages will be buried in quantisation noise. This safety margin is the real value of 24-bit recording. It means you do not need to ride gain as aggressively, and you have more flexibility when editing and mixing.
For anyone recording audio in any context, from studio sessions to voice notes captured on a portable recorder, 24-bit is the standard recommendation. The files are 50 per cent larger than 16-bit equivalents, but storage is cheap relative to the flexibility gained.
32-bit float
A relatively recent development in consumer and prosumer recording equipment, 32-bit floating point audio uses a fundamentally different number representation than standard 16-bit or 24-bit integer audio. The floating-point format can represent an extraordinary dynamic range (roughly 1,528 dB in theory), which makes it effectively impossible to clip the recording at the digital level.
Modern portable recorders from manufacturers like Zoom and Sound Devices now offer 32-bit float recording, and most contemporary DAWs process audio internally in 32-bit float. The practical benefit is significant: even if the input signal is far too hot or far too quiet, the recording captures it without distortion, and levels can be adjusted after the fact with no quality penalty.
For field recording and situations where you cannot monitor levels in real time, 32-bit float is a genuine advancement. It does not change the physics of microphones or preamps, so analogue clipping is still possible, but it removes digital clipping from the equation entirely.
Practical choices
What to record at
Record at 24-bit minimum. The headroom it provides is worth the modest increase in file size, and there is no reason to accept the tighter margins of 16-bit during capture. If your recorder supports 32-bit float, use it for situations where levels are unpredictable.
For sample rate, 48 kHz is the best default for nearly all purposes. It covers both music and video workflows, avoids sample rate conversion in mixed-media projects, and captures the full range of human hearing with margin to spare. Use 44.1 kHz only if you are working exclusively on a music project that will never touch video, and even then, 48 kHz does no harm.
What to mix and edit at
Work at whatever sample rate you recorded at. Converting sample rates mid-project introduces unnecessary processing. Your DAW likely processes internally at 32-bit float or 64-bit float regardless of the source file's bit depth, so the internal processing quality is already higher than your source material.
If your audio project files include recordings at mixed sample rates (common when incorporating library sounds or samples), most DAWs handle real-time conversion transparently. Where possible, match everything to one rate before beginning work.
What to deliver at
This depends on the destination. For music distribution (streaming, downloads), the final master is typically 16-bit/44.1 kHz, dithered down from the higher-resolution mix. For video work, 16-bit or 24-bit at 48 kHz is standard. For broadcast, specifications vary by network and region, but 48 kHz is universal.
When delivering stems or multitracks to a mixing or mastering engineer, match the session's native format, typically 24-bit at the session's sample rate. When archiving completed projects, keep the highest-resolution versions you have. Storage in a cloud workspace costs far less than re-recording.
The high-resolution audio debate
The market for "hi-res audio" has grown considerably, with streaming services like Apple Music, Tidal, and Amazon Music offering lossless and high-resolution streams at sample rates up to 192 kHz and bit depths of 24 bits. The marketing suggests a significant audible improvement over standard CD-quality audio.
The science tells a more measured story. Multiple peer-reviewed studies, including a widely cited meta-analysis published in the Journal of the Audio Engineering Society, have found that listeners cannot reliably distinguish high-resolution audio from CD-quality audio in properly controlled blind tests. The Nyquist theorem guarantees that 44.1 kHz captures everything the human ear can hear, and 16-bit provides more dynamic range than any recording environment or playback situation requires.
This does not mean high-resolution audio is a fraud. Recording and processing at higher resolutions provides real engineering benefits during production. The question is whether those benefits survive the final mastering stage and reach the listener's ears. For the vast majority of people, on the vast majority of playback systems, the answer from controlled research is: probably not.
That said, there is no penalty to listening in high resolution if your storage and bandwidth allow it. The files are larger, and your playback equipment needs to support the format, but you lose nothing by choosing lossless over lossy. The important thing is understanding that the audible improvement over well-encoded 16-bit/44.1 kHz audio, if any, is subtle at best.
File size implications
Sample rate and bit depth both directly affect file size, and the relationship is linear. Doubling the sample rate doubles the file size. Moving from 16-bit to 24-bit increases file size by 50 per cent. These effects multiply: a 24-bit/96 kHz file is three times the size of a 16-bit/44.1 kHz file of the same duration.
For a single track, these differences are manageable. For a session with dozens or hundreds of tracks, the choice of sample rate and bit depth determines whether your project occupies 5 GB or 15 GB on disk. When you factor in backups and versioning, the storage implications are significant. Choosing the right format for archival, such as FLAC instead of WAV, can help manage these demands without sacrificing quality.
Understanding how these settings affect storage is part of evaluating how much cloud storage you need as a working audio professional. A single album project at 24-bit/48 kHz, including all takes, comps, and bounces, can easily run to 20 GB or more before mixing is complete.
A sensible default
If you want a single recommendation to cover most situations: record at 24-bit, 48 kHz, in WAV format. This gives you professional-quality recordings with enough headroom for any editing or mixing you might do, at a file size that is manageable with modern storage. It works for music, podcasts, sound design, film, and video.
When the project is finished, deliver or distribute in whatever format the destination requires, and archive the original high-resolution files. Use smart organisation to keep your originals, working files, and deliverables distinct. The conversion from high-resolution to standard-resolution is simple and lossless in one direction. Going the other way is not possible.
Frequently asked questions
What sample rate should I record at?
48 kHz is the best general-purpose default. It covers both music and video workflows and captures the full range of human hearing. Use 44.1 kHz if your project is exclusively music and will never be paired with video. Higher rates (96 kHz, 192 kHz) are rarely necessary for recording and significantly increase file sizes.
What bit depth should I use for recording?
24-bit is the standard recommendation for all recording. It provides generous headroom for level variations and a lower noise floor than 16-bit, with only a 50 per cent increase in file size. If your recorder offers 32-bit float, use it for unpredictable recording situations like field recording or live events.
Can you hear the difference between 16-bit and 24-bit audio?
In a finished, mastered recording, the difference is not perceptible to listeners under normal conditions. The benefit of 24-bit is during production: it gives engineers more headroom and flexibility when recording and mixing. The final delivered product is typically dithered to 16-bit with no audible penalty.
Is 96 kHz better than 48 kHz for listening?
Controlled blind tests consistently show that listeners cannot reliably tell the difference between 48 kHz and 96 kHz audio. Both capture frequencies well beyond the upper limit of human hearing. 96 kHz may offer marginal benefits during certain types of digital processing, but these advantages do not translate into an audible improvement for the end listener.
What is dithering and why does it matter?
Dithering is the process of adding a tiny amount of noise when reducing bit depth (for example, from 24-bit to 16-bit) to prevent distortion artefacts called quantisation distortion. Proper dithering ensures that the reduction in bit depth is transparent to the listener. It should be applied once, at the final stage of mastering, when converting to the delivery format.
Why is 48 kHz the video standard but 44.1 kHz the music standard?
The 44.1 kHz rate was chosen for CDs in the early 1980s based on the technical constraints of the recording equipment available at the time (it related to the frame rates of video tape machines used for digital audio mastering). The 48 kHz rate was subsequently established as the professional audio and broadcast standard. The difference is historical rather than technical; both rates capture the full range of human hearing.
What does 32-bit float mean on a portable recorder?
32-bit float recording uses a floating-point number system that can represent an enormous dynamic range. In practical terms, it means the digital recording cannot clip. If the signal is too loud, you can reduce the level after recording without any quality loss. This is different from 32-bit integer, which would simply offer a wider fixed dynamic range. 32-bit float recorders still rely on analogue preamps, so the analogue stage can still distort if overdriven.
Do I need high-resolution audio files for transcription?
No. Modern audio transcription tools work well with standard-quality audio. A clear recording at 16-bit/44.1 kHz or even a compressed MP3 will transcribe just as effectively as a high-resolution file. What matters far more is the clarity of the speech, the absence of background noise, and the quality of the microphone.
How do sample rate and bit depth affect streaming quality?
Streaming services transcode uploaded audio to their own formats and quality tiers. Uploading at 24-bit/48 kHz or higher ensures the service has the best possible source material to transcode from. The listener's experience is then determined by the service's encoding settings and the listener's subscription tier, not the original file's specifications.
Should I convert old 16-bit recordings to 24-bit?
Converting a 16-bit file to 24-bit does not add any audio information; it simply pads each sample with additional zero bits. The file gets 50 per cent larger without any improvement in quality. Keep 16-bit recordings in their original format. If you need to process them extensively, your DAW's internal 32-bit float processing will handle them with more than enough precision.