PhD in Psychoacoustic modelling for complex soundscapes

Description of specific topics: You will focus on one of the following:

1.      Development of psychoacoustic models (basic psychoacoustic research):

Sound quality metrics are often used to analyse complex sound scenarios, e.g., for soundscape applications. Sound quality can also affect the health and well-being of people in a given environment. Therefore, it is of the utmost importance that the definition of good sound quality in a given context is as precise as possible. In this regard, psychoacoustic indicators are usually used to develop these metrics.

In recent years, several psychoacoustic standards have been published based on the Sottek Hearing Model: the SHM Loudness, a new approach to time-varying loudness based on a nonlinear combination of partial tonal and noise loudness (as part of the SHM Tonality, standardised in ECMA 418-2) to better account for the fact that the loudness of tonal components, i.e., tonal loudness, may have a stronger influence on loudness perception than the loudness caused by the other components, i.e., noise loudness. In addition, there are standards for psychoacoustic modulation analysis: the SHM Roughness for the assessment of fast modulated sounds (standardised in ECMA 418-2) and the SHM Fluctuation Strength, an adapted model for slow modulated sounds (to be standardised later in 2024).

Other very important psychoacoustic parameters are sharpness and impulsiveness. Existing sharpness models have some drawbacks: for example, they do not adequately account for loudness and temporal effects, and they do not provide sharpness values that linearly correspond to human perception. In addition, there is only a German standard for stationary sounds. For impulsiveness there is currently no standard, but a model based on the Sottek Hearing Model published in 1993.

Improved models based on the recently standardised Sottek Hearing Model should be developed.

Objectives:

• Improved models for sharpness and impulsiveness.
• Evaluation of complex acoustic environments using sound source separation methods:

Acoustic environments often consist of a multitude of different individual sound sources with varying acoustic qualities. Despite this, groups of participants can reliably reach a consensus regarding the most dominant sound sources and the most important qualities of such environments.

When perceiving an acoustic environment, humans intuitively separate different sound sources using both ear signals. Previous studies have shown that sound sources with a clearly perceived direction are often judged differently from non-directional sources. For example, sound sources with annoying qualities are rated as even more annoying when they have a clear direction. Therefore, identifying dominant sources and their directions from a binaural recording can help to better assess the perceived qualities of a given acoustic environment.

Objectives:

• Algorithmic identification of dominant sources and their direction from binaural recordings.
• Separation of dominant sources based on direction for individual auralisation and analysis.
• Extension of both approaches to multi-channel spatial audio (ambisonics) recordings.

The research would suit graduates with a solid understanding and expertise in signal processing and coding. Experience in statistics (or machine learning) and jury testing will be useful.

A little more detail on the projects is as follows:

1. Existing models assume simple sound field situations such as free field or diffuse field. These situations are extremely rare, but an equalization procedure can be applied to binaurally recorded signals that correspond to a simple microphone recording. Normal measurement conditions in rooms or small enclosures such as vehicle interiors will give results that are far from single microphone recordings with different signals for each ear. In this case, equalization is often performed assuming one of the sound fields mentioned above, leading to approximate solutions. A better solution is to use signals measured directly in the ear canal. Here, updated models with new robust reference points are needed: this requires jury tests with human subjects together with measurements in the ear canal and numerical simulations.

2. While psychoacoustic parameters allow a good estimation of the perceived qualities of a sound when analysing single sound sources or time-stable signals, they fail to adequately predict these qualities when an acoustic environment reaches a certain complexity. Therefore, it is important to better understand the different interactions between sources in complex acoustic scenarios in order to better estimate the effect of these scenarios on human perception.

The PhD will begin in January 2025 or September 2025.

The Acoustics Research Centre at the University of Salford is a world-renowned institution with a vibrant research community. You will benefit from: expert supervision from leading academics in acoustics; access to state-of-the-art research facilities; the opportunity to collaborate with researchers from other disciplines, and a supportive, inclusive, and stimulating research environment.

HEAD acoustics has made significant contributions to the field of acoustic and psychoacoustic research. HEAD acoustics was established with a focus on binaural recording and playback technology, sound and vibration analysis, and communication measurement technology. Over the years, the company has grown to become one of the world’s leading suppliers in these fields. HEAD acoustics has been at the forefront of psychoacoustic research, developing new algorithms for psychoacoustic analysis. The Sottek Hearing Model simulates the signal processing of the human auditory system, explaining and describing psychoacoustic effects and basic auditory impressions in detail. This work developed at HEAD acoustics has set the standard in psychoacoustics, helping manufacturers to improve the sound quality of their products.

In Autumn 2025 you will then join the new EPSRC CDT in Sustainable Sound Futures when it launches. The CDT is an unprecedented collaboration for doctoral training between four universities and over 50 project partners. The CDT will provide extensive training that goes far beyond what is normally available for standard PhDs in Acoustics. A mixture of week-long residentials, master classes, theme days and online training will develop technical skills for acoustics (simulation, measurement, machine learning, psychoacoustics, etc.) and key skills for research (project planning, entrepreneurship, public engagement, policy influencing, responsible innovation, etc.). Placements in industry or academic partners will play an important role in ensuring students learn about context and how to create impact.