If you are interested in sound synthesis or audio effects, you may have heard of a term called velvet noise. But what exactly is velvet noise and how can you use it for audio processing? In this blog post, I will explain the basics of velvet noise, its properties, and some examples of how to apply it for different purposes.
What is Velvet Noise?
Velvet noise is a type of sparse random signal that consists of only a few non-zero elements, either +1 or -1, in a given time interval. The name “velvet” comes from the perceptual impression that it sounds smoother than white noise, which has equal power across all frequencies. Velvet noise can be generated by randomly placing impulses with a certain density in a buffer of zeros. The density of the impulses determines the spectral characteristics of the velvet noise. For example, a higher density means more high-frequency content and less low-frequency content.
Why Use Velvet Noise?
Velvet noise has some interesting properties that make it useful for various audio signal processing applications. Some of these properties are:
- It has a flat magnitude spectrum, meaning that it has equal power across all frequencies. This makes it suitable for decorrelation, which is a technique to reduce the similarity between two signals by adding random phase shifts to their frequency components.
- It has a low crest factor, meaning that the ratio between the peak and the average amplitude is small. This makes it suitable for reverberation, which is a technique to simulate the reflections of sound in a room or a space. Velvet noise can create natural-sounding reverberation without causing clipping or distortion.
- It has a low autocorrelation, meaning that the correlation between the signal and its delayed version is small. This makes it suitable for spatialization, which is a technique to create the illusion of sound coming from different directions or locations. Velvet noise can create realistic spatialization effects without causing comb filtering or phasing issues.
- It has a high perceptual sparsity, meaning that it is easy to distinguish from other sounds. This makes it suitable for sound synthesis, which is a technique to create new sounds from scratch or by modifying existing sounds. Velvet noise can create rich and complex sounds with simple algorithms and parameters.
How to Use Velvet Noise?
There are many ways to use velvet noise for audio processing, depending on your goals and preferences. Here are some examples of how to use velvet noise for different purposes:
- To create decorrelation effects, you can simply add velvet noise to your input signal with a certain gain level. You can also use different densities of velvet noise for different frequency bands to create more complex decorrelation effects.
- To create reverberation effects, you can use velvet noise as the input to a feedback delay network (FDN), which is a structure that consists of multiple delay lines connected in a loop with feedback and mixing matrices. You can also use different densities of velvet noise for different delay lines to create more realistic reverberation effects.
- To create spatialization effects, you can use velvet noise as the input to a vector base amplitude panning (VBAP) algorithm, which is a method that calculates the gains for multiple loudspeakers based on the desired direction of the sound source. You can also use different densities of velvet noise for different loudspeakers to create more immersive spatialization effects.
- To create sound synthesis effects, you can use velvet noise as the input to a resonator or a filter bank, which are systems that modify the frequency response of the input signal by emphasizing or attenuating certain frequency bands. You can also use different densities of velvet noise for different resonators or filters to create more diverse sound synthesis effects.
Conclusion
Velvet noise is a type of sparse random signal that has some interesting properties that make it useful for various audio signal processing applications. You can use velvet noise for decorrelation, reverberation, spatialization, and sound synthesis effects with simple and efficient algorithms. Velvet noise can help you create smoother, natural-sounding, realistic, and complex audio effects with minimal computational cost and parameter tuning.