What are HRTFs and how is VisiSonics Revolutionizing Them?

Our ears are unique to us. Yes they help us hear, but they also play a significant role in how we perceive sound. The shape of the outer ear, or pinna, as well as the shape of our heads and torsos – parameters that are unique to each and every one of us – all play a part in coloring the sound that our ears receive. Accordingly, our hearing systems and individual listening experiences become finely tuned to our specific anatomy.

Small differences in the measurements or relationships associated with one’s physiology can have a dramatic effect on sound accuracy and realism.

A Head Related Transfer Function (HRTF) characterizes how an ear receives a sound from a point in space. It is an equation that defines the way sound scatters off a person’s head, shoulders and ears, and ultimately enters the ear canals. Together with room modeling, it is one of the defining elements in an effective virtual spatial audio environment.

The ability to experience spatial audio content through headphones requires a process known as binaural rendering. This allows stereo headphones to create the perception of space and dimensionality through the two stereo channels. Most engines rendering binaural audio rely on a single default HRTF that represents average physical characteristics. It can be considered an HRTF captured from the average human ear. This model, however, does not factor physical differences that can vary person to person, such as the aforementioned parameters: the shape and dimensions of one’s head, ears, or shoulders. Even minor physical differences between one’s anatomy and the default HRTF can result in a very compromised spatial audio listening experience.

The ability to capture a personalized HRTF specific to one’s ears is fundamental to the process of creating a customized spatial audio experience, optimized to that given person’s hearing and sound perception systems.

Incorporating individualized HRTFs into a high fidelity sound engine can significantly improve the perceived quality and realism of binaurally rendered spatial audio. Accordingly, a way to capture personalized HRTF is clearly not only important, but highly desirable.

Traditional methods of measuring individual HRTFs tend to be cumbersome, expensive, and require physical access to the subject. These methods aren’t practical for most, due to the considerable amount of equipment and know-how required to obtain the data. To address these issues, VisiSonics developed a method to extract HRTFs from easy-to-obtain visual information, aided by machine learning and our proprietary database of traditionally acquired HRTFs.

With VisiSonics technology, it is possible to generate an accurate HRTF from individual pictures of the left and right ears, easily taken from a smartphone.

We use a combination of anthropometric feature matching and low frequency “head-and-torso” (HAT) models to create a personalized HRTF. Given a subject’s ear photos and head measurements, the model is used to extract photo landmarks, which are then used to find the closest ear matches from our extensive HRTF database. As a subject’s left and right ears may differ considerably, we process the left and right ears individually. We do not assign significance to whether any given ear is a left or right ear. By doing so, we can match a right ear to a left one, or vice versa, in order to find the best match.

A personalized HRTF is then generated from the matched ears, which is then further tuned to adjust the HRTF to better fit the aforementioned head measurements. In addition, VisiSonics takes personalization to the next level with audiogram technology that enables customization based upon individual listening capabilities.

VisiSonics personalizes sound with our unique, customized HRTF modeling and audiogram measurements.

At VisiSonics, we combine our spatial audio rendering engine and personalization technology to create a fully optimized sound, customized to each individual’s ears. Thus, a few small steps using a smartphone are all it takes to enjoy immersive, custom tailored spatial audio experiences.

We currently make HRTFs available to integrate with all our RealSpace 3D Spatial Audio applications. This includes our embedded DSP solutions, software solutions and our plugins for gaming developers using Wwise, Unity and Unreal.

Contact us today to discuss how you can integrate RealSpace into your next product.