• 検索結果がありません。

Summary

ドキュメント内 東京藝術大学リポジトリ (ページ 87-90)

3.4.4 Conclusion

We have verified the effectiveness of the widening processing proposed in this experiment.

By conducting listening experiments, the correlations between synthesis width and perceived width were investigated. The results showed that for the cello source signal with 1/12-octave bandwidth, perceived width increased with increasing synthesis width, suggesting that under appropriate conditions this method could effectively control the perceived width of a monophonic source in binaural synthesis. Analysis by comparing with stimuli without widening processing and decorrelated references suggested that after processing by the widening method the perceived width was significantly wider than unprocessed one and comparable to the decorrelated stimulus, while the degradations of timbre and spatial quality were less notable than the decorrelation method. However, some limitations were found such as source signal dependency, degradation, and subjective variation. This suggests modifications of parameters, including adjustment regarding source signal characteristics.

The parameter of bandwidth did not show significant effect on the perceived width.

Nevertheless, it could be assumed that how finely frequency components were divided may have influence on the stability of performance and timbre degradation. For both center positions similar results were shown, suggesting that this method could be used on sources from different directions. Further studies are necessary for investigations of the effect of the source direction on widening performance including directions not on the horizontal plane.

3.5 Summary

In this chapter, the presented three experiments investigated a proposed method which aims to produce widths for monophonic sources in binaural synthesis. By dividing mono-phonic sources into octave bands and convolving frequency bands with HRTFs from different directions in a random or deterministic way, stimuli with different synthesis widths were generated. The effect of the widening processing method, and influences of parameters on width perception, were investigated through subjective listening experiments. According to the results, the processing method can successfully produce spatial widened sources in binaural synthesis. It has been found that several parameters of the processing have important influences on the perceived source width.

First, the distribution methods, which determine what direction each frequency band is distributed to, have a major influence on the localization of source width. Since usually there is a large variation in energy among different frequency bands of source signals, how to distribute the energy of the signal uniformly to the range of intended synthesis width is a major challenge to be solved for this processing method. If it can be achieved, not only can

localization shifted from the intended direction be avoided, but the spatially quality of the source width may also be improved.

This brings up the idea that how finely frequency bands were divided,i.e., the bandwidth, is also a crucial factor to the performance of source widening. A narrower bandwidth was found to able to ensure a more stable performance of source widening effect. The narrowest bandwidth investigated in this study is 1/12 octave bands. It may be reasonable to suppose that dividing the frequency bands more finely could further improve the performance, but timbre degradation may also occur.

Different directions of the centers of the source width were examined to investigate the effectiveness of widening processing on source at various directions. The results show similar effect of widening process, although only centers of 0° and 15° azimuth were investigated.

The effectiveness of source widening was only significant with sufficiently large synthesis width (generally larger than 40°). In addition, individual differences and a dependency of performance on characteristics of source signals were found, suggesting that further investigation and improvement of the widening processing are needed.

The purpose of developing a source widening effect for binaural synthesis is to provide a method to create and control the width of sound objects in binaural reproduction, with an assumption that producing sound objects with spatial extent, which is closer to the complex auditory events we usually experience in natural auditory environments, can achieve better spatial impression. The widening effect can be used in a wide range of applications such as binaural reproduction of object-based audio, binaural mixing consoles, and virtual reality audio. Therefore, to demonstrate the feasibility of applying the widening effect in audio production, an experiment was conducted and is presented in the next chapter.

Chapter 4

Spatial impression of source widening effect for binaural audio production

4.1 Introduction

In Chapter 3, three experiments were conducted to investigate the effect of the proposed source widening processing method. The results of subjective listening experiments showed that source signals could have wider perceived widths after processed by widening effect. In this study, to investigate the effect of the widening processing on the spatial impression in the practical application of audio production, the widening process method was implemented as a VST plugin for real-time processing in binaural audio reproduction. VST (Virtual Studio Technology) is an audio plugin software interface which can accommodate digital audio signal processing of synthesizers and effects in digital audio workstation (DAW). The VST plugin was used for sound effects mixing with widening effect applied to source signals of sound effects. Instead of using noises and anechoic recordings as in experiments described in Chapter 3, the source signals were chosen considering the practical audio production situation. The aim of this experiment was to investigate whether the synthesis of sound source widths could provide a better spatial impression of the work.1

1The results of this experiment were also published inH. Su, A. Marui, and T. Kamekawa, “Spatial Impression of Source Widening Effect for Binaural Audio Production,” presented at the Audio Engineering Society Conference: 2018 AES Int. Conf. on Spatial Reproduction-Aesthetics and Science (2018).

ドキュメント内 東京藝術大学リポジトリ (ページ 87-90)

関連したドキュメント