Misleading Industry Jargon 1:

Bit Depth and Sample Rates

You may have heard or seen the PCM resolution of audio recordings stated or proudly posted on CD covers as marketing fluff, but like all misleading advertising, the consumer, and sometimes even marketers, have little idea what the truth is about the jargon. Here are two of the most widely misunderstood terms in audio. There is a popular belief in the Audiophile industry that the higher the number of each, the better. True in some ways, yes, but rather pointless in others.

In simple terms, with PCM audio (Pulse Code Modulation), an audio signal is sampled a certain number of times per second, and the voltage of that signal is recorded at each sample point. If certain criteria is met, the computer can capture a continuous waveform and reconstruct it perfectly from each sample using “simple” math. Turning a continuously variable, or analog waveform, into a digital chart of samples, i.e. a .wav file, and back into a continuously variable analog waveform. The popular myth is that a digital waveform is constructed from some kind of stepped, jagged, staircase wave that degrades the audio quality. This is simply not true and perpetuates another myth that recording at a higher sample rate will lead to a smoother digital wave form. In fact it will not.

Sample Rates: Typical sample rates for audio signals are 44.1kHz (CD), 48kHz(video standard), and multiples such as 88.2 and 96kHz or higher. Here is the only important truth about sample rates that should matter to the listener: the higher the sample rate, the higher the audio frequency that can be recorded. Since sample theory states that a minimum of two samples must be used to reconstruct a perfect waveform, the highest audio frequency that can be recorded in digital is a little less than half of the sample rate. So the highest pitch is ~20kHz for CD audio, which is the extreme limit for human hearing. Even if a recording was sampled at 96kHz, achieving over 40kHz of frequency information, all that high frequency information will be lost once converted to CD or video standards.

This leads to one of the most frequent arguments in the digital audio industry over whether it is even worth it to use the extra drive space and effort to record high sample rates when a) humans cannot hear it, b) the most popular playback mediums of CD and mp3 cannot replicate it, c) most microphones cannot capture it, and d) most instruments cannot even produce frequencies that high. There are even good arguments that feeding speakers higher frequencies than they are designed to handle will cause more distortion than if the frequency were limited in the first place. There are a couple good uses for high sample rates, including “oversampling” for digital peak limiting, but for the most part, the listener will not notice any difference, and you should not be concerned with the quality of recordings sampled at 44.1kHz. They should sound the same as anything higher.

Bit Depth: The bit depth refers to the number of steps between the loudest and softest signals that can be recorded. At CD resolution of 16 bits there are exactly 65,536 steps (2^16). This translates to a “noise floor” of 96dB* . This was the resolution introduced early on in digital commercial recordings. Since then the recording bit depth has graduated from 20 bits of the ADAT machines to modern 24 bit resolution. This may not seem like a huge leap, after all it is only half again as many bits at CD quality, but trust the math: 24bit resolution allows 16,777,216 steps between the noise floor and the loudest possible signal and a theoretical noise floor of 144dB**.

Theoretical because it is far lower than the analog capabilities of even the highest quality mastering converter. This is 256 times the resolution of “CD quality”, and what this means is that the engineer can record at much lower volumes to preserve the dynamics of the signal, negating the need for destructive dynamic processing at the input. This is a very good thing for us on the production end, but after the signal has been manipulated for proper listening levels, 16bits is more than adequate. If you are wondering where .mp3 and other compressed formats come in, that is a topic for another article. For now be assured that even the simplest modern conversion machines can achieve phenomenal results. The real difference in a $10, and a $5000 converter lies in the quality of design and construction, and the transparency of the analog components. A good professional converter will have a noise floor of 114 to ~130 dB. Analog tapes and records have noise floor around 40dB for reference.

*Noise floor refers to the lowest point below digital clipping that a bit depth can achive. 96dB means 96 decibles below the loudest possible sound. At this point no audio information can be recorded, and gain applied to the signal will be filled with noise as a placeholder.
**Also note that in digital, 6dB equates to a doubling of perceived volume.