Tom Kelly
- Apr 23, 2020
- 3 min read

17. Sample Rate, Bit Depth, and MP3 File Conversion Listening Examples

As we start to wrap up this long series on frequency, we work our way towards the natural conclusion, export settings. While sample rate and bit depth can be and are recording settings and are decided at the beginning of the session, we also have options when bouncing down (see also: exporting) our sessions. The same goes for what bitrate mp3 to convert our audio to. Each of these settings affects frequency either super directly or mildly indirectly so we're going to focus a bit on HOW our voices are converted and encoded into a digital signal, as well as how it's converted back into an analog signal.

JOIN THE FUN ON PATREON!

Subscribe at

Apple Podcasts | Google Podcasts | Overcast | Spotify | Pocketcasts

Here's some things we are defining in this episode.

Sample Rate

When an analog signal (your voice) is converted into a digital signal, our ADAC (analog to digital/ digital to analog converter) is taking snapshots of our analog signal at regular intervals. The common numbers that come to mind are 44.1k and 48k. What this means is 44,100 times per second, our converters are taking a "picture" of our amplitude and assigning that a value in either positive or negative. The higher the sample rate, the more accurately we can recreate this sine wave above.

Nyquist Theorem

The Nyquist Theorem, sometimes also called the Nyquist-Shannon sampling theorem states that in order to accurately reproduce an analog sound digitally, we must have a sample rate that's equal to or greater than double the highest frequency we are trying to sample. The audible range of human hearing is (theoretically) 20Hz-20kHz. To accurately convert and reproduce frequencies up to 20kHz, we must have a sample rate of AT LEAST 40kHz. Why 44.1? Adding a little extra for error and the slope of a low pass filter.

Aliasing

If we use a sample rate that's too small to accurately capture and reproduce the frequencies present in the recording, we get a phenomenal called aliasing. This is an inaccurate capturing and recreation of a complex waveform due to too few samples. The photo above will do a better job of explaining this than I could ever do in words. It's also cray helpful to HEAR aliasing. Just listen to the episode. Trust me...

Bit Depth

When we are capturing amplitude readings at our chosen sample rate, it is rounded to the nearest available amplitude value. The higher the bit rate, the more available values to be rounded to, resulting in a more accurate recreation of the analog signal. I like to think of this as pixels in a photograph. The more available pixels, the sharper the detail of each. Our number of available values goes up EXPONENTIALLY as we raise our bit rate.

16 bit has 65,536 possible integer values per sample

24 bit has 16,777,216 possible integer values per sample

32 bit has 4,294,967,296 possible integer values per sample