A Basic Overview of Dithering

 In TSM Studio Blog
  • The application of dither, although rooted in a relatively simple principle, can seem daunting to engineers that have never been tasked with delivering master files at a lower bit-depth. Fortunately, applying dither under the proper circumstances is clear-cut and easily comprehensible.
  • First, you’ll need to understand what bit-depth is before you can grasp the concept of dithering. Bit-depth, or bit-resolution, refers to a numerical value that describes the number of bits of information present in any given sample. A higher bit-depth correlates to a higher dynamic range within a digital audio file. Real-world examples of bit-depth include CD players and Blue-Ray players. CD players utilize 16 bits per sample and Blu-Ray players support up to 24 bits per sample.
  • In conjunction with understanding bit-depth, it’s imperative that you understand quantization errors before we can discuss dither. Quantization errors are a result of the difference between a continuous analog signal’s value and the closest available digital value at each sampling instant. Because analog signal values are infinite and digital signal values are finite, when the analog value falls between two digital values, the nearest digital value will represent the analog signal, resulting in a very slight error that introduces noise.
  • Now that you have a basic understanding of bit-depth and quantization errors, you’re now able to comprehend dithering and its application. Dither is a low-level form of noise or erroneous signal that’s introduced into a digital audio file when converting from a higher bit-resolution to a lower bit-resolution. The primary purpose of dithering is to minimize the inherent quantization error that occurs when converting digital audio files.
  • You may be asking yourself, “why would I want to introduce noise to my recordings?”. The reason for adding this subtle noise to your digital audio file is because it trades the unpleasant sound of truncation distortion for noise that rounds off the errors that occur during digital file conversion
  • Your next question is likely, “what’s truncation distortion?”. Don’t worry, it’s just another technical term relative to the quantization errors we discussed earlier. Truncation distortion is a form of distortion that occurs as a result of reduced bit-resolution by cutting off the remaining numbers to fit into a lower bit-depth. For example, when reducing bit-depth from 24 bit-resolution to 16 bit-resolution, truncation distortion will present itself if dithering isn’t applied to round off quantization errors. When you reduce your bit-depth from 24-bit to 16-bit, you’re reducing the number of numerical values that are able to measure the dynamic range of a given sample.
  • To grasp this concept more fluidly, imagine asking both an entry level music student and a classically trained jazz musician to explain the concept of harmony. The jazz musician would have a much more descriptive response about the concept of harmony in comparison to the musician who is just learning the basics of music theory. Think of the 24 bit-resolution as the more descriptive, complex interpretation given by the jazz musician and the 16 bit-resolution as the less detailed interpretation from the novice musician. When going from 24 bit-resolution to 16 bit-resolution, a digital audio file’s dynamic range becomes less detailed and transparent, as would the explanation of a classically trained musician in comparison to a student.
  • With a fundamental understanding of dither, we need to discuss when to apply it. Oftentimes, dithering is left to the mastering engineer as they typically deal with the technical specifications associated with preparing files for final delivery across various mediums. Nonetheless, as engineers it’s our job to be familiar with these adaptive methods. By following these guidelines, you’ll know when it’s appropriate to dither and when it’s best to save it for later down the line.
  • If you’re not down-sampling, don’t dither. If your audio is being converted from 24-bit/48kHz to 16-bit/44.1kHz, dithering will be necessary. If your files are remaining at 24-bit/48kHz, there’s no need to dither because no quantization or truncation errors are going to be introduced.
  • Only dither your rendered audio if it’s the final version. If an audio file isn’t mastered yet, the mastering engineer will take care of the dithering as their final master will need to undergo the process.