Dithering in audio is a technique used during the final stages of digital audio processing, particularly when reducing the bit depth of an audio file. It involves adding a small amount of random noise to the audio signal. This might seem counterintuitive, but dithering helps to mask or minimize the distortion due to truncation that can occur when the bit depth is reduced.
Why Dithering is Important:
Bit Depth Reduction: When converting audio from a higher bit depth to a lower one, quantization errors can occur. These errors manifest as distortion or unwanted noise, especially in the quieter parts of the audio. Dithering helps to smooth out these errors.
Noise Shaping: The random noise added through dithering can be shaped in a way that it is less noticeable to the human ear, often pushing the noise into frequencies that are less audible.
How it Works:
When you reduce the bit depth, some of the finer details of the audio are lost because the number of possible values (or levels) that the audio can represent decreases. Dithering introduces a low level of noise that can randomize the quantization errors, making them less correlated with the original audio signal and therefore less perceptible.
Types of Dithering:
Rectangular Dither: The simplest form of dither, where noise is added uniformly across the signal.
Triangular Dither: Triangular dither is commonly used in audio applications due to its smoother sound.
Noise Shaped Dither: Focuses the added noise into higher frequencies, where it’s less noticeable to the human ear. It should only be used at the final stage if necessary.
Practical Use:
Dithering is typically applied as the last step of the music production process when exporting audio to a lower bit depth, such as when preparing a track for CD production (which typically uses 16-bit audio). It’s important to note that dithering should only be applied once; multiple applications can increase noise and degrade the sound quality.
As of today, most DAWs and OS use floating-point bit depth which increases the calculation accuracy during processing in the digital domain. So, at each stage of the music production process, when a file is exported to 16 or 24 bits integer, dithering should be applied. If the export remains to a floating-point bit depth, either 32 or 64, dithering should not be applied.
If an audio file has an integer bit depth and is imported in a DAW, every tiny modifications such as a gain change or EQ will result in a calculation in floating-point. So, if the new file export is an integer bit depth, dithering should be applied.
Conclusion :
In summary, dithering helps maintain audio quality when reducing bit depth by minimizing unwanted artifacts like distortion. Best practices recommend avoiding dithering during file transfers between the recording, mixing, and mastering stages, reserving it for the final mastering step. If exporting to an integer bit depth, such as 16 or 24 bits, it’s recommended to use TPDF dithering for optimal results. Remember, any audio file imported into a DAW that undergoes modifications will require dithering again if exported at an integer bit depth.