Digital Audio

Your DAW asks you for a sample rate, a bit depth, a buffer size and a file format. These settings decide how faithfully your computer captures sound, how much room you have before distortion, and how big your files are. This lesson explains each one in plain words and tells you exactly what to pick.

Analog vs Digital

Sound in the air is a smooth, continuous wave of pressure (see What is Sound?). A microphone turns that pressure into a smooth, continuously changing voltage. That's an analog signal: it can take any value at any moment.

A computer can only store numbers. To record sound, it measures the voltage many thousands of times per second and writes each measurement down as a number. That's a digital signal.

• The part that measures is the ADC (analog-to-digital converter).
• The part that turns numbers back into voltage for your speakers is the DAC (digital-to-analog converter).

Both live inside your audio interface (and inside every phone and laptop).

Two settings decide how accurate that conversion is: sample rate (how often you measure) and bit depth (how precisely you measure).

Sample Rate

Each measurement is called a sample. The sample rate is how many samples are taken per second, measured in Hz or kHz.

At 44.1 kHz, the computer takes 44,100 measurements every second for each channel.

Drag the "Samples per cycle" slider down. With only a few samples per wave, the stored copy (coral) becomes a rough staircase that no longer follows the original (grey). With many samples, the copy follows the curve closely.

The Nyquist rule

The Nyquist theorem says a sample rate can capture frequencies up to half of itself. That limit is called the Nyquist frequency.

If a frequency above the Nyquist limit sneaks in, it doesn't just disappear. It "folds back" and shows up as a wrong, lower frequency. This is called aliasing, and it sounds harsh and out of tune. Converters use filters to block it, and some plugins offer oversampling to reduce it inside the DAW.

Common sample rates

Bit Depth

Bit depth is how many possible values each sample can have. More bits means finer steps between the quietest and loudest levels.

• 16-bit: 65,536 possible levels.
• 24-bit: about 16.8 million possible levels.

When a measurement falls between two levels, it gets rounded to the nearest one. That rounding error is called quantization error, and you hear it as low-level noise. Go back to the demo, press Play and lower the bit depth: at 2–4 bits the tone becomes gritty and distorted.

Dynamic range: about 6 dB per bit

Dynamic range is the gap between the loudest sound a format can hold and its noise floor. Each bit adds roughly 6 dB.

What is 32-bit float?

32-bit floating point stores numbers in a way that can go far above 0 dBFS and far below the noise floor. Most DAWs mix internally in 32-bit float (or 64-bit). That's why a channel meter can go into the red without the file being ruined, as long as you turn it down before it reaches the master output or a fixed-point file.

File Formats

Audio files come in two families: uncompressed / lossless, which keep every sample, and lossy, which throw away detail to make files smaller.

Lossy formats use psychoacoustics to remove sounds you're unlikely to notice. At high bitrates they sound very close to the original, but each time you re-encode, more is lost. Learn more in Psychoacoustics.

Mono vs Stereo

A mono file has one channel. Played on two speakers, the same signal comes out of both, so it sounds like it sits in the middle. A stereo file has two channels, left and right, which can be different. That difference creates width.

• Record single sources (one voice, one guitar, one mic) in mono. You can pan and widen them later.
• Synths, pads, drum loops and room recordings are often stereo.
• Kick, bass and lead vocals are usually kept mono and centered in a mix.

Stereo files are twice the size of mono files. See Panning & Stereo for how to use width in a mix.

Latency and the Buffer

Your computer can't process audio one sample at a time; it works in chunks called the buffer. Collecting a chunk takes time, which creates latency: a delay between playing something and hearing it.

Under about 10 ms round trip, most people can play and sing comfortably. Above roughly 20 ms, the delay starts to feel noticeably off. Use a small buffer (64–128) when recording and a large one (512–1024) when mixing. Many interfaces also offer direct monitoring, which sends the input straight to your headphones with no computer delay at all.

Clipping and Headroom

In digital audio, 0 dBFS is the absolute ceiling. Any sample that tries to go higher gets chopped flat. That's clipping, and it causes harsh, crackly distortion. Unlike analog gear, which distorts gradually, digital clipping is sudden and ugly.

Headroom is the space between your loudest peak and 0 dBFS. Leaving headroom gives you a safety margin for loud moments and room for processing later.

With 24-bit recording there's no benefit to recording "hot". The noise floor is so low that peaks at -12 dBFS lose nothing, and you're safe from a sudden loud note clipping. You'll go deeper in Mixing & Gain Staging and Loudness & LUFS.