Final Report Summary - TEPESS (Technologies and psychophysics of spatial sound)
The project has worked in the different subfields of spatial sound. The psychoacoustics of spatial sound addresses the resolution how well humans are able to hear the spatial characteristics of sound scene; in which directions, how far, and of what size are the sources; and how one perceives the space where the sound is performed. The technologies for spatial sound consist different methods to record, store, and reproduce sound with spatial characteristics in controlled manner.
This project focused on generic audio format, a sound format which would enable delivery of spatial sound in such a manner, that as good spatial audio quality would be perceived in different multi-channel loudspeaker listening setups, and with headphone listening. The generic audio format addressed in this study consists of few audio channels accompanied with time-frequency-domain metadata describing the most prominent directions and diffuseness parameters of sound field. A handful of methods to convert different existing multichannel audio formats, and from different microphone array inputs to generic audio format were developed, and the reproduction methods were optimized for different listening setups. The weakest step in associated processing is decorrelation, which is a potential source of audible artifacts. In this work we identified the signal-space conditions when such artifacts emerge, and suggested three methods to overcome the artifacts.
The studies on psychophysics of spatial sound concentrated on perception of spatially complex sound scenarios, the cases where sound source is distributed into a large region, and the task of the human is to report the spatial distribution of the source, or to report the sounds heard from different parts of the source. The studies revealed that humans are relatively weak in perception of spatial resolution, and also that in often the similarity of sounds coming from different directions makes the listeners to group them to a single auditory object, though in some interesting cases the sounds are segregated into several streams. Such resolution issues in human spatial hearing are then used in optimization of spatial audio techniques.
In accurate headphone reproduction of sounds with headphones, the response should be equalized as well as possible. In practise this would mean to measure the response of headphones to the ear drum, as well as the response from a loudspeaker to the ear drum, and then the headphone response could be equalized according to the measurements. In practise this is hazardous, and subject to many errors due to microphone positioning and other factors. We investigated the possibility to measure the responses from the opening of the ear canal using a miniature pressure-velocity probe, and could show that the measurement is possible, and that it has better tolerance to microphone placement.
A spinoff result from the project is a technique, where a powerful laser can be used to measure the acoustical effect of a space to sound, e.g the reverberation of a concert hall. The laser is focused to a single point in space, which causes a miniature plasma explosion emitting an impulsive sound with very high level. The impulse can be used to measure the impulse response of the space, or the scale model of a space. The impulse has perfectly omnidirectional directional pattern, no mass, and very high level, which makes it optimal for such measurements.
Another spinoff result from the project is a new type of acoustical beamformer, which could be utilized to capture sound with narrow directional patterns from microphones mounted on a small device, such as video camera or other mobile device. The microphone signals are first decomposed into a new set of signals with spherical harmonic directional patterns, and in second phase the cross-coherence of the spherical-harmonic signals is utilized to modulate one of the signals in time-frequency domain. This provides a robust and effective beam also at low frequencies.
This project focused on generic audio format, a sound format which would enable delivery of spatial sound in such a manner, that as good spatial audio quality would be perceived in different multi-channel loudspeaker listening setups, and with headphone listening. The generic audio format addressed in this study consists of few audio channels accompanied with time-frequency-domain metadata describing the most prominent directions and diffuseness parameters of sound field. A handful of methods to convert different existing multichannel audio formats, and from different microphone array inputs to generic audio format were developed, and the reproduction methods were optimized for different listening setups. The weakest step in associated processing is decorrelation, which is a potential source of audible artifacts. In this work we identified the signal-space conditions when such artifacts emerge, and suggested three methods to overcome the artifacts.
The studies on psychophysics of spatial sound concentrated on perception of spatially complex sound scenarios, the cases where sound source is distributed into a large region, and the task of the human is to report the spatial distribution of the source, or to report the sounds heard from different parts of the source. The studies revealed that humans are relatively weak in perception of spatial resolution, and also that in often the similarity of sounds coming from different directions makes the listeners to group them to a single auditory object, though in some interesting cases the sounds are segregated into several streams. Such resolution issues in human spatial hearing are then used in optimization of spatial audio techniques.
In accurate headphone reproduction of sounds with headphones, the response should be equalized as well as possible. In practise this would mean to measure the response of headphones to the ear drum, as well as the response from a loudspeaker to the ear drum, and then the headphone response could be equalized according to the measurements. In practise this is hazardous, and subject to many errors due to microphone positioning and other factors. We investigated the possibility to measure the responses from the opening of the ear canal using a miniature pressure-velocity probe, and could show that the measurement is possible, and that it has better tolerance to microphone placement.
A spinoff result from the project is a technique, where a powerful laser can be used to measure the acoustical effect of a space to sound, e.g the reverberation of a concert hall. The laser is focused to a single point in space, which causes a miniature plasma explosion emitting an impulsive sound with very high level. The impulse can be used to measure the impulse response of the space, or the scale model of a space. The impulse has perfectly omnidirectional directional pattern, no mass, and very high level, which makes it optimal for such measurements.
Another spinoff result from the project is a new type of acoustical beamformer, which could be utilized to capture sound with narrow directional patterns from microphones mounted on a small device, such as video camera or other mobile device. The microphone signals are first decomposed into a new set of signals with spherical harmonic directional patterns, and in second phase the cross-coherence of the spherical-harmonic signals is utilized to modulate one of the signals in time-frequency domain. This provides a robust and effective beam also at low frequencies.