Recording and Production

Friday, October 18, 9:00 am — 10:15 am (1E11)

Chair:
Tomasz Zernicki, Zylia sp. z o.o. - Poznan, Poland

EB1-1 Recording and Mixing of Classical Music Using Non-Adjacent Spherical Microphone Arrays and Audio Source Separation Algorithms—Eduardo Patricio, Zylia Sp. z o.o. - Poznan, Poland; Mateusz Skrok, Zylia Sp. z o.o. - Poznan, Poland; Tomasz Zernicki, Zylia sp. z o.o. - Poznan, Poland
The authors present a novel approach to recording classical music, making use of non-adjacent 3rd order Ambisonics microphone arrays. The flexible combination of source separated signals with varied degrees of beamforming focus enable independent levels control, while maintaining the spatial coherence and reverberation qualities of the recorded spaces. The non-coincidental arriving locations of multiple arrays allow for post-production manipulations without disrupting the inherent classical music logic that values the overall sound as opposed to individual single sound sources. In addition, this method employs portable and lightweight equipment to record decorrelated signals, which can be mixed in surround formats with enhanced sense of depth.

EB1-2 Exploring Preference for Multitrack Mixes Using Statistical Analysis of MIR and Textual Features—Joseph Colonel, Queen Mary University of London - London, UK; Joshua D. Reiss, Queen Mary University of London - London, UK
We investigate listener preference in multitrack music production using the Mix Evaluation Dataset, comprised of 184 mixes across 19 songs. Features are extracted from verses and choruses of stereo mixdowns. Each observation is associated with an average listener preference rating and standard deviation of preference ratings. Principal component analysis is performed to analyze how mixes vary within the feature space. We demonstrate that virtually no correlation is found between the embedded features and either average preference or standard deviation of preference. We instead propose using principal component projections as a semantic embedding space by associating each observation with listener comments from the Mix Evaluation Dataset. Initial results disagree with simple descriptions such as “width” or “loudness” for principal component axes.

EB1-3 Machine Learning Multitrack Gain Mixing of Drums—Dave Moffat, Queen Mary University London - London, UK; Mark Sandler, Queen Mary University of London - London, UK
There is a body of work in the field of intelligent music production covering a range of specific audio effects. However, there is a distinct lack of any purely machine learning approaches to automatic mixing. This could be due to a lack of suitable data. This paper presents an approach to used human produced audio mixes, along with their source multitrack, to produce the set of mix parameters. The focus will be entirely on the gain mixing of audio drum tracks. Using existing reverse engineering of music production gain parameters, a target mix gain parameter is identified, and these results are fed into a number of machine learning algorithms, along with audio feature vectors of each audio track. This allow for a machine learning prediction approach to audio gain mixing. A random forest approach is taken to perform a multiple output prediction. The prediction results of the random forest approach are then compared to a number of other published automatic gain mixing approaches. The results demonstrate that the random forest gain mixing approach performs similarly to that of a human engineer and outperforms the existing gain mixing approaches.

EB1-4 Why Microphone Arrays Are Not Better than Single-Diaphragm Microphones with Regard to Their Single Channel Output Quality—Helmut Wittek, SCHOEPS Mikrofone GmbH - Karlsruhe, Germany; Hannes Dieterle, SCHOEPS Mikrofone GmbH - Karlsruhe, Germany
A comparison of the directional characteristics of single-diaphragm vs multi-microphone arrays is performed on the basis of frequency response and polar diagram measurements. The simple underlying question was: Is a conventional first-order pressure-gradient microphone better than an M/S array or an Ambisonics microphone regarding the quality of their individual outputs? The study reveals significant differences and a clear superiority of single-diaphragm microphones regarding the smoothness of on- and off-axis curves which is believed to highly correlate with timbral fidelity. Array microphones, on the other hand, can potentially create variable patterns and a higher order directivity. [Presentation only; not in E-Library]

EB1-5 Predicting Objective Difficulty in Peak Identification Task of Technical Ear Training—Atsushi Marui, Tokyo University of the Arts - Tokyo, Japan; Toru Kamekawa, Tokyo University of the Arts - Adachi-ku, Tokyo, Japan
Technical ear training is a method to improve the ability to focus on a speci?c sound attribute and to communicate using the vocabularies and units shared in Audio Engineering. In designing the successful course in a sound engineers’ educational institution, it is essential to have a gradual increase in the task dif?culty. In this e-Brief, the authors investigated creating a predictive model of objective dif?culty for a given music excerpt when it is used in a peak identi?cation task of technical ear training. The models consisting of six or seven acoustic features, including statistics on attack transients and power spectrum, showed overall better results.