|SPHEROVOX PENTAHORN OMNIVOX / DODECAHEDRON CORNER FOLDED HORN ACOUSTIC MEASUREMENTS LATEST DEVELOPMENT ABOUT US CONTACS|
MUSIC ACADEMY HALL ACOUSTIC MEASUREMENTS
BNR STUDIO ONE ACOUSTIC MEASUREMENTS
SYMPHONY ORCHESTRA HALL 'BULGARIA' ACOUSTIC MEASUREMENTS
SOFIA OPERA HOUSE ACOUSTIC MEASUREMENTS
NATIONAL THEATRE 'IVAN VAZOV' ACOUSTIC MEASUREMENTS
RESULTS OF THE MEASUREMENTS
AUDIO ENGINEERING SOCIETY (AES)
PRESENTATION [PowerPoint] [Acrobat]
OMNIVOX VERSUS SPHEROVOX COMPARISONS
NEWS The Audio Engineering Society Comes to Los Angeles for 137th International AES Convention
AES137 to feature busy itinerary of numerous track sessions, including Recording and Production, the Project Studio Expo, Networked Audio, and more AES136 Berlin Highlights Project Studio Expo and Special Sound for Picture Programming
Celebrating the success of the 136th Audio Engineering Society Convention, which took place April 26 - 29, 2014 at the Estrel Hotel and Convention Center in Berlin, Germany, the AES continues to forge new territory with its successful tracks and programs at recent Conventions. AES Showcases Latest Audio Innovations and Best Practices at 136th International Convention in Berlin
Successful Technical Program and Exhibition draw high attendance and positive response from attendees and exhibitors alike
COMPARATIVE ACOUSTIC MEASUREMENTS: SPHERICAL SOUND SOURCE VS. DODECAHEDRON
For acoustic measurements exclusively dodecahedron type omnidirectional sound sources are used currently for both impulse
response measurements in architectural acoustics and sound isolation measurements in building acoustics. As new originally
designed, compression driver based spherical sound source for acoustic measurement recently became available, a logical
step was to make comparative comprehensive acoustic measurements of some well accepted rooms for speech and music using
both sources under the same conditions.
The aim of the measurements is not to differentiate and compare acoustic parameters and room qualities, but rather to compare the results obtained by implementing two completely different types of sound sources. Dodecahedron, which is correctly called Omni-Directional, as far as it radiates in 12 different directions with its 12-loudspeakers mounted on a “spherical” surface, is very well investigated and is exclusively used for acoustic measurements.
The second sound source operational principle is quite the opposite – horn loaded compression drivers are concentrated in the center of the sphere and radiate the same field in any direction. That’s why it is logical to call it spherical sound source. To better differentiate and explain the comparative measurements, the spherical sound source would be named Spherovox for short, and the dodecahedron would be named Omnivox. This name would not validate the current results and comments for all dodecahedrons on the market, but rather only for our own built dodecahedron for the comparative purposes. Omnivox is a typical Dodecahedron as far as it has 38–cm distance between any two opposing penta-hedron surfaces, and has 6.5 inch loudspeakers implemented. This dodecahedron sound source was designed with the best loudspeaker component available on the market – 6.5 inch co-axial drivers BMS-6CN160 with dual ND magnetic structures including mid-size compression drivers. With these components and with active electronics and high power amplification 500W/250W implemented, this dodecahedron source might be unbeatable among its equals, and it is very challenging to compete with.
The Spherovox sound source composes 2-pcs of dual membrane large format compression drivers BMS-4599-SPH modified, loaded by a common radially expanding horn, and two 15-inch Co-axial loudspeakers BMS-15ND-SPH. Amplification is done by a common 1200W amplifier for both LF drivers and a 250W for their HF Co-axial drivers, while the 2-mid-high frequency dual membrane compression drivers are powered by a 500W bridged amplifier. SELECTED ROOMS FOR COMPARATIVE MEASUREMENTS Five halls for speech and music performances have been carefully selected for comparative acoustic measurements.
Three well accepted concert halls with different dimensions have been investigated, including Sofia Philharmonic Orchestra Hall-Bulgaria, The National Academy of Music Hall in Sofia and Studio One of the Bulgaria National Radio, which is often used as a concert hall for classical and jazz performances. The first two halls are equipped with full size pipe organ installations.
Two more halls for speech and music performance are chosen for comparative sound source performance evaluation, including the main stage of the Ivan Vazov National Theatre and the Sofia Opera and Ballet Hall. MUSIC ACADEMY HALL ACOUSTIC MEASUREMENTS MUSIC ACADEMY HALL SOFIA Figure 1. MUSIC ACADEMY HALL SOFIA ACOUSTIC PARAMETERS » Results of measurements BNR STUDIO ONE ACOUSTIC MEASUREMENTS BNR STUDIO ONE Figure 2. BNR STUDIO ONE ACOUSTIC PARAMETERS » Results of measurements SYMPHONY ORCHESTRA HALL 'BULGARIA' ACOUSTIC MEASUREMENTS SYMPHONY ORCHESTRA HALL 'BULGARIA'
Figure 3. SYMPHONY ORCHESTRA HALL 'BULGARIA' ACOUSTIC PARAMETERS » Results of measurements SOFIA OPERA HOUSE ACOUSTIC PARAMETERS SOFIA OPERA HOUSE Figure 4. SOFIA OPERA HOUSE ACOUSTIC PARAMETERS » Results of measurements NATIONAL THEATRE 'IVAN VAZOV' ACOUSTIC MEASUREMENTS NATIONAL THEATRE 'IVAN VAZOV' Figure 5. NATIONAL THEATRE 'IVAN VAZOV' ACOUSTIC PARAMETERS » Results of measurements MEASURING PROCEDURE AND TECHNIQUE In all the rooms two sound source positions are used, except for the opera house where the orchestra pit was opened and a third source position was placed just in the middle of it. The first source position is put on the room axis of symmetry at about 3m backward from the front of the stage. The second sound source positions are placed 2-4 m apart from the first and are unsymmetrically offset at about the same distance. The alternative sound source positions were exactly the same, at 1.5m height of their acoustic centers in all the houses.
Six microphone positions are placed on each room’s main floor, which are randomly distributed onto one symmetrical half of the rooms. When balconies are available, two more microphone positions are placed on each of them, on the same hemi-axial side. Thus 6-Mic positions are available for the two smaller concert halls, 8-Mic positions are recorded in the much bigger Sofia Philharmonic orchestra hall, where a single balcony is most often used, and 10-Mic positions are used in the two theatres.
For impulse response recording, stereo microphone settings are used implementing both MS and Binaural technique. For acoustic measurements, MS consists of Omni-directional and Figure-8 Bi-directional diagram microphones, contrary to the sound recording practice, where M-microphone (Mid) in MS setup is actually cardioid. Artificial head Neumann KU81i was used as a binaural microphone pair.
For Total and Lateral energy measurements a number of mics are checked as Omnis and as Figure-8. Bruel & Kaer ½ inch diffuse field equalized measuring mic Type 4134 was used as a reference for comparisons with a number of AKG451E-C2 samples and one of them was picked to be used for current comparative measurements. Its frequency response was within 0.5 dB as close to that of the measuring B&K mic all the way from 50Hz to 10kHz.
A number of Figure-8 microphones were evaluated to pick up the most suitable one, including 1-inch dual membrane Neumann mics with switchable diagrams. Schoeps Figure-8 mic type CMC6U-MK8, picked up for the current measurements exhibited the most symmetrical polar pattern at high frequency and was the smallest.
The external microphone preamplifier had a fixed gain of 20 dB for all measuring positions in all measuring rooms for both its inputs, which was valid for both types of settings – MS and Binaural. Six to 10 dB Omni to Figure-8 microphone sensitivity difference was software equalized. This was done for further conveniently absolute SPL difference comparisons of IR recorded into different halls.
Both sound sources were excited by the same, about 10 sec long, sine-sweep signal with the same level and the same power amplification in all rooms and each of them giving exactly the same individual sound power levels in all rooms, with constant frequency spectrum.
Even though the spherical sound source - Spherovox is capable of producing more than 10dB higher SPL at mid-frequency, and even 6dB more for low and high ends (at the same 6-dB amplifiers clip margin), its level was reduced to much that of the dodecahedron Omnivox for better and easier acoustic parameter comparisons.
Spherovox SWL response was measured to be flat, as designed, down to 50 Hz (-3dB) and could be effectively used from 63 Hz octave center frequency, which is visible on some diffuse field plots. At that frequency dodecahedron would have more than 10 dB less capability, which makes it effective from 100 Hz (-3dB) upwards. That’s why we haven’t made any attempts to extract acoustic parameters for the 63Hz octave band, and respectively to put parameters for that band from spherical sound source IRs only.
Five sound source rotational steps are used for every microphone position in any of the rooms evaluated. On-axis room direction is marked as Zero (0), (1) and (2) directions are 15 and 30 deg respectively clockwise, while 3 and 4 directions are -15 and -30 deg i.e. counter clockwise.
For dodecahedron, on-axis Zero direction corresponds to one of the symmetry axis coinciding with the so-called “vertical” plain of measurements . For Spherovox the zero position is irrelevant because of the rotational symmetry – just 15 deg step matters to some extent.
Comprehensive 1/3 octave free field directivity measurements were carried out on both sound sources. A single vertical polar pattern measurement is available for any frequency band of the spherical sound source, which gives all the data needed to construct precise spherical radiation balloon diagram, eventually a precise Sound Power Level graph. For dodecahedron loudspeaker however, even though four different planes of such 1/3 octave polar pattern measurements are performed, they were very far from being enough to construct a precise balloon diagram, respectively a precise SWL frequency response. That’s why Strength G related acoustic parameters are excluded from the results presented. ACOUSTIC PARAMETERS OF UPPERMOST IMPORTANCE A number of different sets of acoustic parameters are published as being the most important for objective room acoustic evaluation and comparison of different rooms, subjectively evaluated by famous composers, conductors and musicians and ranked by some prominent acousticians including – Beranek . Usually such parameter sets are ranked as of importance of their ingredients, which might differ in order of importance to some extent for big concert halls, smaller chamber orchestra halls and opera theaters.
For spacious room acoustic evaluation, however, almost exclusively only two types of stereo microphone measurement techniques are used for recording and extracting all room’s parameters from stereo IR. One of the techniques records Lateral (mono channel combined mean square magnitude of lateral Left-Right reflections) IR with Figure 8 microphone, usually on the right channel, and the total Impulse Response due to the sounds coming from all direction with an omni-directional microphone is recorded on the left channel. The second stereo technique uses a binaural artificial head with two microphones installed into the pinnae, to evaluate cross correlation between the two ears’ signals - the more correlation the less spaciousness, IACC (Inter Aural Cross-Correlation) equals to 1 in free space. To better answer the logic - the higher the better, Beranek has introduced the parameter Binaural Quality Index BQI equals to 1-IАСС.
Although both spacious parameters could not distinguish between Left or Right Lateral contribution to the extracted parameters, Binaural measurement contains in its IR itself both Left and Right channels recorded and clearly distinguishable, as well as measurable, to determine the actual sound (reflection) content.
If the sound source directivity unevenness is responsible for differences between left/right lateral reflection proportions during stepped source rotation in a given position, this will bring less impact on both Lateral Fraction and IACC (and it even might be cancelled altogether), than on individual frequency responses truncated from essential parts of the Binaural stereo IR.
With the current comparative acoustic measurements only the standard simple stereo microphone measurements were carried out, trying to obtain high quality 24 bit IR with maximum dynamic range possible, for any further calculations of additional parameters. RESULTS OF THE MEASUREMENTS
Reverberation time RT30 and EDT are presented in a common characteristic, which includes Bass Ratio (BR) defined in many articles including  as Low to Mid RT30 Ratio and Brilliance, defined as High to Mid RT30 Ratio. Both ratios are dimensionless, but for simplicity it is presented on the common vertical axis where reverberation is read in seconds. The two ratio parameters are positioned at their corresponding octaves as lines between them.
Clearance parameters, defined for music and speech as dB ratio of the first 80msec or respectively 50msec from the IR to the total sound energy, are presented on a common graph, which include for completeness two shorter initial time intervals namely 35msec and 7msec.
Lateral Fraction and IACCE (Inter Aural Cross Correlation Coefficient - early) parameters are presented on a common chart on which BQI (Binaural Quality Index) defined from Beranek  as (1-IACCe) is used instead. The third parameter shown on the same graph is STI (Speech Transmission Index). Its average value for the corresponding room is plotted in the middle between the two upper and lower curves marking the standard deviation of the parameter throughout any source-mic stepped pair positions. Spherovox obtained STI is plotted between the first two octave center frequencies, immediately followed by STI plotted for Omnivox by dashed lines between the following two octaves. This allows clear and convenient comparison of room intelligibility parameters. All these three parameters might be strange to be seen on the same chart, but as far as they are all dimensionless and could have possible values between 0 and 1, it is at least mathematically acceptable to be placed together, which would save a lot of explanations and graphics.
On all common charts for both sources, solid lines are used for the spherical sound source (Spherovox), and the dodecahedron (Omnivox) graphs are plotted with dashed lines. This permits the acoustic parameters measured by both sound sources to be easily distinguished and directly compared on a single chart.
In some of the plots showing spacious parameters LF and BQI dodecahedron dashed curves represent respective parameters averaged in two groups - directions 0-1-3 and directions 2-4. In the first group all spacious parameters for all source-mic pair positions from 3 IR taken 15-deg apart are averaged, which is room on-axis and +/- 15 deg steps off axis. The second group contains two directions 2-4, which are 60 deg apart, or +/- 30 deg on both sides of the room axis. As can be read from the plots, the two grouped characteristics differ substantially from one another. By averaging parameters taken from more rotational stepped direction dodecahedron IR, the spacious parameters seem to go closer to those of the spherical sound source. Still it is a bit of gambling, because some rotational step multiples might point all their individual directional radiation towards the mic (or wall-mic), with either source directivity diagram deep or peek. For our particular axis of rotation, the example for such troubled step is 32 deg. All 32 deg steps have a chance to radiate only peeks towards the mic, no matter how many they are - 5, 10 or more, and if offset at 16 deg clockwise (or counterclockwise), these same set steps might position the source to radiate with their deeps towards the receiver for all steps.
By increasing the number of rotational steps it looks as if spacious parameters become more evenly frequency distributed. It is however at least arguable if it is correct to average spacious parameters at all.
|Print Top Send|