A recently proposed auditory model is examined using simulated binaural masking level difference ... more A recently proposed auditory model is examined using simulated binaural masking level difference (BMLD) and dichotic pitch (Huggins and binaural edge pitch) stimuli. The model is based on calculating the instantaneous interaural level difference, i.e. the difference between the left and right ear neural signals. The model output produces pronounced maximum at the signal frequency with BMLD stimulus. Both dichotic pitch stimuli produce a notable maxima at the pitch frequencies. Although the model can thus be interpreted to predict known psychoacoustical results, an exact quantitative comparison with the model responses and data from BMLD and binaural pitch experiments is not performed. Rather, this paper serves as a "proof of concept".
Journal of The Audio Engineering Society, May 1, 2009
In teleconferencing application of Directional Audio Coding, the transmitted data consists of mon... more In teleconferencing application of Directional Audio Coding, the transmitted data consists of monophonic audio signal and directional metadata measured in frequency bands depending on time. In reproduction, each frequency channel of the signal is reproduced to ...
In multi-channel systems, if a subwoofer is not used, low frequencies may also be reproduced with... more In multi-channel systems, if a subwoofer is not used, low frequencies may also be reproduced with multiple loudspeakers. When the active loudspeakers are relatively close to each other, they act like a single transducer and form a directivity pattern. This causes unwanted effects in the systems' low-frequency responses, depending on the listening location. The use of wave field synthesis (WFS) at low frequencies was hypothesized to reduce unwanted effects caused by directivity and the room modes. To study this hypothesis, different reproduction techniques were studied objectively and subjectively at the frequency region of 50-100 Hz in a listening room. The systems included one-channel mono, three-channel and eight-channel amplitude panning, and wave field synthesis. The data from measurements and listening tests shows that the systems have distinct perceptual differences. The WFS method was determined be a promising alternative for low-frequency reproduction, since it uses the maximum number of loudspeakers available in the setup and provides a fairly uniform sound pressure distribution over a large listening area.
Journal of The Audio Engineering Society, May 26, 2016
Immersive audio experiences (3D audio) are an important element of next-generation audio entertai... more Immersive audio experiences (3D audio) are an important element of next-generation audio entertainment systems. This paper presents joint object coding techniques that enable the delivery of object-based immersive audio content (e.g. Dolby Atmos) at low bit rates. This is achieved by conveying a multi-channel downmix of the immersive content using perceptual audio coding algorithms together with parametric side information that enables the reconstruction of the audio objects from the downmix in the decoder. An advanced joint object coding tool is part of the AC-4 system recently standardized by ETSI. Joint object coding is also used in a backwards compatible extension of the Dolby Digital Plus system. Listening test results illustrate the performance of joint object coding in these two applications.
This paper presents a novel computational auditory model inspired by psychoacoustic and neurophys... more This paper presents a novel computational auditory model inspired by psychoacoustic and neurophysiological findings. The model utilizes the instantaneous difference between the two ear signals to predict spatial hearing cues that humans perceive. The main focus is on the interaural coherence cue, i.e. the perceived similarity between the waveforms of the ear signals. Simulations show that the proposed model is capable of predicting known psychoacoustical results.
Recent neurophysical studies suggest that binaural decoding is based on count comparison in cases... more Recent neurophysical studies suggest that binaural decoding is based on count comparison in cases of both ITD and ILD decoding. In such mechanisms, the neural signals are stronger in the auditory pathways leading to the ipsilateral hemisphere when a signal is presented earlier, or with higher level, to the contralateral ear. This paper describes a simple computational model implementing binaural cue decoding based on countcomparison principles. In the model, ITD and ILD are decoded using separate mechanisms, inspired by the functions of Medial Superior Olive and the Lateral Superior Olive found in the mammal brainstem. It is also assumed, based on psychoacoustic data, that ITD decoding is sluggish, and ILD decoding is faster. It is shown that the proposed mechanisms decode static ITD and ILD similarly as humans do. In addition to ITD and ILD, humans can also detect interaural coherence, or the similarity between ear signals. Interaural coherence is in the model decoded as the range of temporal variations of ILD.
In recent years, the task of Automatic Music Transcription (AMT), whereby various attributes of m... more In recent years, the task of Automatic Music Transcription (AMT), whereby various attributes of music notes are estimated from audio, has received increasing attention. At the same time, the related task of Multi-Pitch Estimation (MPE) remains a challenging but necessary component of almost all AMT approaches, even if only implicitly. In the context of AMT, pitch information is typically quantized to the nominal pitches of the Western music scale. Even in more general contexts, MPE systems typically produce pitch predictions with some degree of quantization. In certain applications of AMT, such as Guitar Tablature Transcription (GTT), it is more meaningful to estimate continuous-valued pitch contours. Guitar tablature has the capacity to represent various playing techniques, some of which involve pitch modulation. Contemporary approaches to AMT do not adequately address pitch modulation, and offer only less quantization at the expense of more model complexity. In this paper, we present a GTT formulation that estimates continuous-valued pitch contours, grouping them according to their string and fret of origin. We demonstrate that for this task, the proposed method significantly improves the resolution of MPE and simultaneously yields tablature estimation results competitive with baseline models.
ABSTRACT A binaural audio synthesis system based on sinusoidal modeling is proposed for spatial, ... more ABSTRACT A binaural audio synthesis system based on sinusoidal modeling is proposed for spatial, low-bitrate audio coding utilized for example in teleconference applications. The system transmits monaural sinusoidal parameters of a downmix signal, from which the left and right binaural signals are synthesized according to the directional metadata at the receiver. Typical sinusoidal synthesis methods, as well as the effectiveness of a monaural frequency masking model, are evaluated in binaural context. Furthermore, a method for binaural noise residual synthesis and efficiency improvements for HRTF parameter acquisition are suggested. Tests utilizing speech signals indicate that sinusoidal modeling is an attractive technique for applications such as the proposed one.
Sinusoidal modeling of audio requires the model parameters to be selected by analyzing the origin... more Sinusoidal modeling of audio requires the model parameters to be selected by analyzing the original signal spectrum. This paper proposes two improvements in sinusoidal selection by considering how psychoacoustic masking curves can be calculated using a top-down strategy in certain situations. First, a non-iterative component selection method to be used in combination with an added residual signal is presented. Tests indicate computational gain and quality increase when the method is used with a noise-synthesized residual. Secondly, the estimation of the masking curve in binaural listening when signals are panned is considered. Tests show that knowledge of the degree of panning is bene cial when heavy panning is applied to simultaneously rendered audio object signals.
This paper focuses on a comparative statistical analysis of the performance of VoIP calls under v... more This paper focuses on a comparative statistical analysis of the performance of VoIP calls under various situations, namely, during a handover and under different background traffic conditions at a wireless access point (AP). Using empiricalbased measurements, it demonstrates that these network conditions exhibit distinct statistical behaviour, in terms of SNR, packet losses and end-to-end delays, and thus, impact the VoIP user quality in a different manner. The analysis shows that both network conditions and codec type, as well as their interaction, have a significant effect on the PESQ MOS values. Moreover, it indicates statistically highly significant differences between the estimations of the PESQ and E-model. Finally, it highlights the benefits of the packet loss concealment of the AMR 12.2kb/s under these network conditions.
A recently proposed auditory model is examined using simulated binaural masking level difference ... more A recently proposed auditory model is examined using simulated binaural masking level difference (BMLD) and dichotic pitch (Huggins and binaural edge pitch) stimuli. The model is based on calculating the instantaneous interaural level difference, i.e. the difference between the left and right ear neural signals. The model output produces pronounced maximum at the signal frequency with BMLD stimulus. Both dichotic pitch stimuli produce a notable maxima at the pitch frequencies. Although the model can thus be interpreted to predict known psychoacoustical results, an exact quantitative comparison with the model responses and data from BMLD and binaural pitch experiments is not performed. Rather, this paper serves as a "proof of concept".
Journal of The Audio Engineering Society, May 1, 2009
In teleconferencing application of Directional Audio Coding, the transmitted data consists of mon... more In teleconferencing application of Directional Audio Coding, the transmitted data consists of monophonic audio signal and directional metadata measured in frequency bands depending on time. In reproduction, each frequency channel of the signal is reproduced to ...
In multi-channel systems, if a subwoofer is not used, low frequencies may also be reproduced with... more In multi-channel systems, if a subwoofer is not used, low frequencies may also be reproduced with multiple loudspeakers. When the active loudspeakers are relatively close to each other, they act like a single transducer and form a directivity pattern. This causes unwanted effects in the systems' low-frequency responses, depending on the listening location. The use of wave field synthesis (WFS) at low frequencies was hypothesized to reduce unwanted effects caused by directivity and the room modes. To study this hypothesis, different reproduction techniques were studied objectively and subjectively at the frequency region of 50-100 Hz in a listening room. The systems included one-channel mono, three-channel and eight-channel amplitude panning, and wave field synthesis. The data from measurements and listening tests shows that the systems have distinct perceptual differences. The WFS method was determined be a promising alternative for low-frequency reproduction, since it uses the maximum number of loudspeakers available in the setup and provides a fairly uniform sound pressure distribution over a large listening area.
Journal of The Audio Engineering Society, May 26, 2016
Immersive audio experiences (3D audio) are an important element of next-generation audio entertai... more Immersive audio experiences (3D audio) are an important element of next-generation audio entertainment systems. This paper presents joint object coding techniques that enable the delivery of object-based immersive audio content (e.g. Dolby Atmos) at low bit rates. This is achieved by conveying a multi-channel downmix of the immersive content using perceptual audio coding algorithms together with parametric side information that enables the reconstruction of the audio objects from the downmix in the decoder. An advanced joint object coding tool is part of the AC-4 system recently standardized by ETSI. Joint object coding is also used in a backwards compatible extension of the Dolby Digital Plus system. Listening test results illustrate the performance of joint object coding in these two applications.
This paper presents a novel computational auditory model inspired by psychoacoustic and neurophys... more This paper presents a novel computational auditory model inspired by psychoacoustic and neurophysiological findings. The model utilizes the instantaneous difference between the two ear signals to predict spatial hearing cues that humans perceive. The main focus is on the interaural coherence cue, i.e. the perceived similarity between the waveforms of the ear signals. Simulations show that the proposed model is capable of predicting known psychoacoustical results.
Recent neurophysical studies suggest that binaural decoding is based on count comparison in cases... more Recent neurophysical studies suggest that binaural decoding is based on count comparison in cases of both ITD and ILD decoding. In such mechanisms, the neural signals are stronger in the auditory pathways leading to the ipsilateral hemisphere when a signal is presented earlier, or with higher level, to the contralateral ear. This paper describes a simple computational model implementing binaural cue decoding based on countcomparison principles. In the model, ITD and ILD are decoded using separate mechanisms, inspired by the functions of Medial Superior Olive and the Lateral Superior Olive found in the mammal brainstem. It is also assumed, based on psychoacoustic data, that ITD decoding is sluggish, and ILD decoding is faster. It is shown that the proposed mechanisms decode static ITD and ILD similarly as humans do. In addition to ITD and ILD, humans can also detect interaural coherence, or the similarity between ear signals. Interaural coherence is in the model decoded as the range of temporal variations of ILD.
In recent years, the task of Automatic Music Transcription (AMT), whereby various attributes of m... more In recent years, the task of Automatic Music Transcription (AMT), whereby various attributes of music notes are estimated from audio, has received increasing attention. At the same time, the related task of Multi-Pitch Estimation (MPE) remains a challenging but necessary component of almost all AMT approaches, even if only implicitly. In the context of AMT, pitch information is typically quantized to the nominal pitches of the Western music scale. Even in more general contexts, MPE systems typically produce pitch predictions with some degree of quantization. In certain applications of AMT, such as Guitar Tablature Transcription (GTT), it is more meaningful to estimate continuous-valued pitch contours. Guitar tablature has the capacity to represent various playing techniques, some of which involve pitch modulation. Contemporary approaches to AMT do not adequately address pitch modulation, and offer only less quantization at the expense of more model complexity. In this paper, we present a GTT formulation that estimates continuous-valued pitch contours, grouping them according to their string and fret of origin. We demonstrate that for this task, the proposed method significantly improves the resolution of MPE and simultaneously yields tablature estimation results competitive with baseline models.
ABSTRACT A binaural audio synthesis system based on sinusoidal modeling is proposed for spatial, ... more ABSTRACT A binaural audio synthesis system based on sinusoidal modeling is proposed for spatial, low-bitrate audio coding utilized for example in teleconference applications. The system transmits monaural sinusoidal parameters of a downmix signal, from which the left and right binaural signals are synthesized according to the directional metadata at the receiver. Typical sinusoidal synthesis methods, as well as the effectiveness of a monaural frequency masking model, are evaluated in binaural context. Furthermore, a method for binaural noise residual synthesis and efficiency improvements for HRTF parameter acquisition are suggested. Tests utilizing speech signals indicate that sinusoidal modeling is an attractive technique for applications such as the proposed one.
Sinusoidal modeling of audio requires the model parameters to be selected by analyzing the origin... more Sinusoidal modeling of audio requires the model parameters to be selected by analyzing the original signal spectrum. This paper proposes two improvements in sinusoidal selection by considering how psychoacoustic masking curves can be calculated using a top-down strategy in certain situations. First, a non-iterative component selection method to be used in combination with an added residual signal is presented. Tests indicate computational gain and quality increase when the method is used with a noise-synthesized residual. Secondly, the estimation of the masking curve in binaural listening when signals are panned is considered. Tests show that knowledge of the degree of panning is bene cial when heavy panning is applied to simultaneously rendered audio object signals.
This paper focuses on a comparative statistical analysis of the performance of VoIP calls under v... more This paper focuses on a comparative statistical analysis of the performance of VoIP calls under various situations, namely, during a handover and under different background traffic conditions at a wireless access point (AP). Using empiricalbased measurements, it demonstrates that these network conditions exhibit distinct statistical behaviour, in terms of SNR, packet losses and end-to-end delays, and thus, impact the VoIP user quality in a different manner. The analysis shows that both network conditions and codec type, as well as their interaction, have a significant effect on the PESQ MOS values. Moreover, it indicates statistically highly significant differences between the estimations of the PESQ and E-model. Finally, it highlights the benefits of the packet loss concealment of the AMR 12.2kb/s under these network conditions.
Uploads
Papers by Toni Hirvonen