Projects per year
Abstract
In everyday environments, listeners perceive sound sources as externalized. In listening conditions where the spatial cues that are relevant for externalization are not represented correctly, such as when listening through headphones or hearing aids, a degraded perception of externalization may occur. In this thesis, the spatial cues that arise from a combined effect of filtering due to the head, torso, and pinna and the acoustic environment were analysed and the impact of such cues for the perception of externalization in different frequency regions was investigated. Distant sound sources were simulated via headphones using individualized binaural room impulse responses (BRIRs).
An investigation of the influence of spectral content of a sound source on externalization showed that effective externalization cues are present across the entire frequency range. The fluctuation of interaural level differences (ILDs) that occurs in reverberant environments was altered via modifications of the signal envelope in the left and right ear. It was found that the dynamic ILDs had an effect on externalization for broadband and highpass filtered speech, while no effect was found for lowpass filtered speech. Moreover, the compression of low frequency ITD fluctuations did not influence externalization.
Further, the influence of binaural and monaural cues from reverberation was investigated. It was found that monaural reverberation cues were sufficient for the externalization of a lateral source, whereas a frontal source required an increased amount of binaural cues from reflections in order to attain convincingly externalized sound images. It was concluded that the disparity in the interaural cues of the direct sound and the reverberation was important as the interaction of the two played a role for the perception of externalization. Moreover, similar binaural effects of reverberation were found at low- and high frequencies.
The performance of a multi-microphone noise reduction algorithm designed to preserve binaural cues in hearing aids was investigated in conjunction with a voice activity detector (VAD) for noise estimation. Intelligibility weighted improvements in signal-to-noise ratio (SNR) of 6 dB and 18 dB were found for diffuse multi-talker babble noise and speech shaped directional noise, respectively, at input SNRs close to the speech reception threshold (SRT) of hearing impaired listeners.
Overall, this work contributes to the understanding of the auditory processing of spatial cues that are important for externalization in reverberant environments and may have implications for hearing instrument signal processing.
An investigation of the influence of spectral content of a sound source on externalization showed that effective externalization cues are present across the entire frequency range. The fluctuation of interaural level differences (ILDs) that occurs in reverberant environments was altered via modifications of the signal envelope in the left and right ear. It was found that the dynamic ILDs had an effect on externalization for broadband and highpass filtered speech, while no effect was found for lowpass filtered speech. Moreover, the compression of low frequency ITD fluctuations did not influence externalization.
Further, the influence of binaural and monaural cues from reverberation was investigated. It was found that monaural reverberation cues were sufficient for the externalization of a lateral source, whereas a frontal source required an increased amount of binaural cues from reflections in order to attain convincingly externalized sound images. It was concluded that the disparity in the interaural cues of the direct sound and the reverberation was important as the interaction of the two played a role for the perception of externalization. Moreover, similar binaural effects of reverberation were found at low- and high frequencies.
The performance of a multi-microphone noise reduction algorithm designed to preserve binaural cues in hearing aids was investigated in conjunction with a voice activity detector (VAD) for noise estimation. Intelligibility weighted improvements in signal-to-noise ratio (SNR) of 6 dB and 18 dB were found for diffuse multi-talker babble noise and speech shaped directional noise, respectively, at input SNRs close to the speech reception threshold (SRT) of hearing impaired listeners.
Overall, this work contributes to the understanding of the auditory processing of spatial cues that are important for externalization in reverberant environments and may have implications for hearing instrument signal processing.
Original language | English |
---|
Place of Publication | Kgs. Lyngby |
---|---|
Publisher | Technical University of Denmark |
Number of pages | 105 |
Publication status | Published - 2014 |
Fingerprint
Dive into the research topics of 'Human Sound Externalization in Reverberant Environments'. Together they form a unique fingerprint.Projects
- 1 Finished
-
Distance perception in impaired and aided-impaired hearing
Catic, J. (PhD Student), Dau, T. (Main Supervisor), Santurette, S. (Supervisor), Marozeau, J. (Examiner), van de Par, S. (Examiner) & Seeber, B. (Examiner)
Technical University of Denmark
15/01/2010 → 26/01/2015
Project: PhD