Every master ends as a digital file. Understanding how digital audio works - and what each format needs - lets you deliver masters that keep all the quality you worked for.
PCM: capturing sound
Digital audio uses pulse code modulation (PCM). An analogue-to-digital converter measures the waveform many thousands of times per second and stores each measurement as a number. Two settings define PCM audio: sample rate and bit depth.
Sample rate
The sample rate is how many measurements are taken per second.
- 44.1 kHz - the CD standard, and still common for music.
- 48 kHz - the video standard, increasingly common for music too.
- 88.2, 96 kHz and higher - used for recording and processing, and for high-resolution releases.
The Nyquist-Shannon theorem says that to capture a frequency you need a sample rate of more than twice that frequency. Human hearing extends to about 20 kHz, so 44.1 kHz - capturing up to about 22 kHz - covers the full audible range. Lower sample rates cannot reproduce the highest frequencies.
Bit depth
Bit depth is how precisely each sample's level is measured. It sets the dynamic range and the digital noise floor. Each bit adds roughly 6 dB of dynamic range:
- 8-bit - about 48 dB. Early video game consoles.
- 12-bit - about 72 dB. Classic samplers such as the E-mu SP-1200 and Akai MPC60 - part of their gritty character.
- 16-bit - about 96 dB. The CD standard.
- 24-bit - about 144 dB in theory - far beyond what real converters and rooms achieve. The standard for recording, mixing and mastering.
- 32-bit float - used inside DAWs; effectively impossible to clip internally.
A higher signal-to-noise ratio is better: keep your signal comfortably above the noise floor. Lower bit depths are sometimes used deliberately as an effect, for lo-fi character.
Dither
When you reduce bit depth - for example from 24-bit to 16-bit for a CD master - the quieter details are rounded off, creating a harsh distortion called quantisation error. Dither adds a tiny amount of noise that randomises this error, turning it into a smooth, far less audible noise.
- Dither only once, at the very end, when reducing bit depth.
- Do not dither when exporting at the same bit depth, or at 32-bit float.
- Noise-shaped dither pushes the noise into frequencies where we hear it least.
Sample-rate conversion
If you mastered at a higher sample rate than the delivery format, convert at the end with a high-quality converter. Avoid unnecessary conversions in the middle of the chain.
Delivery formats
- Streaming and digital distribution - usually 24-bit or 16-bit WAV at 44.1 kHz or higher, according to your distributor's requirements. Many now accept and prefer 24-bit.
- CD - 16-bit, 44.1 kHz, dithered, often with a DDP image for the manufacturer.
- Vinyl - a separate master is usually best: less extreme loudness, controlled sibilance and high frequencies, and mono low end. Talk to the cutting engineer.
- Lossy files (MP3, AAC) - for previews and promos only, encoded from the final master at a high bitrate.
- Hi-res - 24-bit at 96 kHz or higher, for platforms and stores that support it.
Naming and versions
Name every file clearly - artist, title, version, sample rate and bit depth. Deliver the versions the release needs: main, clean, instrumental, TV mix (without lead vocal), a cappella.
Do this now
- Check your distributor's audio requirements and write them down.
- Export a master at 24-bit/44.1 kHz with no dither, and a 16-bit version with dither. Compare the quietest fade-out of each on headphones.
- Create a file-naming template for your deliverables.
Adapted, with thanks, from the lesson notes of J.M. Brooklyn Audio Mastering.