Digital Audio
1970s onwardsDigital audio is one idea applied repeatedly: measure the sound often enough, and accurately enough, and the numbers hold everything a listener can hear. Almost every argument about digital sound is really an argument about "enough".
Sampling
Take a measurement of the waveform many thousands of times a second. Each measurement is a sample; play them back in order at the same rate and the wave is reconstructed.
How often is often enough? The governing result is that you must sample at more than twice the highest frequency you want to capture. Human hearing tops out around 20 kHz, so a rate above 40 kHz suffices — which is why the CD's 44.1 kHz is not arbitrary.
A very common misunderstanding is that the output is a staircase of steps. It is not. Given a properly band-limited signal there is exactly one waveform that passes through those points, and reconstruction recovers it — the steps are an artefact of how the process is usually drawn, not of how it works.
Bit depth
Bit depth is how precisely each individual sample is measured. More bits means finer gradations between the quietest and loudest representable sound.
Bit depth therefore controls dynamic range and the noise floor — not, as is often assumed, frequency response or "detail". Sixteen bits, as used on CD, spans a range wider than most listening rooms are quiet enough to reveal.
Higher depths matter most in production, where repeated processing accumulates rounding errors and headroom is needed before anything is mixed down.
Lossless and lossy
Lossless compression — FLAC and ALAC among them — makes the file smaller with no discarded information at all, exactly like zipping a document. Decode it and you have the original bit for bit.
Lossy compression — MP3, AAC — makes files much smaller by permanently discarding parts of the signal judged inaudible. It is far more efficient, and it is one-way.
For listening, a good lossy encode at a high bitrate is very hard to distinguish from lossless. For archiving, only lossless makes sense, because you cannot know what a future format will need.
Why it still gets argued about
Some of the disagreement is real: early converters genuinely were poor, and bad mastering is bad regardless of format. Much of the rest is a mastering argument wearing a format argument's clothes — the same album is often mastered differently for vinyl and for digital, and people compare the masters while believing they are comparing the media.
Names that mattered
Listed because of what they did, not as recommendations. Velorex Music did not make any of these.
Did you know?
- The CD's 44.1 kHz rate follows from needing more than twice the ~20 kHz limit of human hearing.
- Bit depth sets dynamic range and noise floor — not frequency response.
- The "staircase" picture of digital audio is a drawing convention, not what a reconstructed signal looks like.
- Lossless compression is genuinely lossless: decode a FLAC and you get the original bits back exactly.
Remember this?
Digital audio's nostalgia is oddly domestic: ripping a CD collection one disc at a time, correcting track names by hand, and the particular satisfaction of a library where every album finally had the right artwork.
Sources & further reading
- Library of Congress — MP3 format description
- IEEE ETHW — Compact Discs (CDs)
- Philips Research — The history of the CD
Written from these sources, not copied from them. External links open in a new tab.