What are the pros and cons of different types of dithering algorithms and noise shaping curves?

1 What is dithering?

Dithering is the process of adding a small amount of random noise to your audio signal before you convert it to a lower bit depth. This noise helps to mask the quantization errors that may cause audible distortion or artifacts in your audio. Dithering also preserves the dynamic range and the frequency response of your audio, especially in the low-level and high-frequency regions. Dithering is essential when you master your audio for CD, MP3, or other formats that use 16-bit or lower bit depth.

2 What is noise shaping?

Noise shaping is the process of manipulating the frequency spectrum of the dither noise. This means that you can make the noise more or less audible in certain frequency ranges, depending on your preference and the characteristics of your audio. Noise shaping can help you reduce the perceived noise level of your audio, or enhance the clarity and detail of your audio. Noise shaping is often combined with dithering to achieve the best results.

3 Types of dithering algorithms

Dithering algorithms vary in the amount and distribution of noise they add to audio. Rectangular dither adds a uniform amount of noise across the frequency spectrum, but has the lowest effectiveness in masking quantization errors. Triangular dither adds a triangular-shaped noise with a higher probability of being close to zero than rectangular dither, and has slightly higher noise level and effectiveness in masking quantization errors. Gaussian dither adds a bell-shaped noise with an even higher probability of being close to zero than triangular dither, and has a higher noise level and effectiveness in masking quantization errors. Lastly, shaped dither adds a noise shaped according to a specific curve that minimizes the noise level in the audible range and maximizes it in the inaudible range, making it the most effective in masking quantization errors but also having the highest noise level.

4 Types of noise shaping curves

Different types of noise shaping curves can be applied to the dither noise, varying in the amount and direction of the frequency manipulation. For example, no noise shaping results in a flat and uniform noise spectrum across all frequencies, which may be suitable for audio such as speech or ambient sounds. Low-pass noise shaping attenuates the high-frequency components of the dither noise while preserving or boosting the low-frequency components, which may be suitable for audio that needs more warmth and depth such as acoustic or classical music. High-pass noise shaping attenuates the low-frequency components of the dither noise while preserving or boosting the high-frequency components, which may be suitable for audio that requires more clarity and detail such as electronic or rock music. Additionally, custom noise shaping allows users to adjust the frequency manipulation according to their own preference and the characteristics of their audio, which may be suitable for experimental or artistic music.

5 How to choose the best dithering and noise shaping algorithm

There is no definitive answer to which dithering and noise shaping algorithm is the best for your audio mastering; it depends on various factors such as the original and target bit depth and sample rate of your audio, the genre and style of your music, the listening environment and device of your audience, and your personal taste and preference. However, some general guidelines may help you make a better decision. Always use dithering when you reduce the bit depth of your audio, unless you are sure that no quantization errors will occur or that they will not be audible. Consider using noise shaping when you want to reduce the perceived noise level of your audio or enhance its sonic qualities. Choose a dithering algorithm that matches the level of noise reduction and effectiveness that you need, such as rectangular dither for minimal noise reduction or shaped dither for maximal noise reduction. Additionally, select a noise shaping curve that matches the frequency characteristics that you want to emphasize or de-emphasize in your audio; for example, use low-pass noise shaping for more warmth and depth or high-pass noise shaping for more clarity and detail. Ultimately, experiment with different combinations of dithering and noise shaping algorithms, then trust your ears to choose the one that sounds the best.

6 Here’s what else to consider

This is a space to share examples, stories, or insights that don’t fit into any of the previous sections. What else would you like to add?