Can transparent recording be improved by the study of a recording space’s clarity?

This paper has been written for SAE Institute UK, for the purpose of a BSc Major Project.

Abstract:

This paper presents a study of clarity in multiple transparent recordings. With the help of modern acoustic measurements, which further-on inform transparent recording decisions, we are looking to understand more about sonic clarity in recording scenarios, performed inside two different churches. The measurements were focused on the Clarity Index C-80 curve and the Definition Index D-50 curve. The results helped us better understand clarity inside the two churches we recorded in. Listening comparisons between microphones positioned based on the results and microphones not positioned based on the results were conducted. The positioning of both the microphones and the performers proved to be altering to the clarity of the recording. The comparisons indeed proved that the measurements helped us improve the clarity of the recording. We also noticed that clarity and reverberation do not see eye to eye, which means that the more you have from one the less you have from the other.

Introduction:

“For just as musical instruments are brought to perfection of clearness in the sound of their strings by means of bronze plates or horn sounding boards, so the ancients devised methods of increasing the power of the voice in theatres through the application of the science of harmony.” — Vitruvius (Vitruvius Pollio and Morris Hicky Morgan (trans.), 'Book V: Chapter III', Vitruvius, the Ten Books on Architecture (1914), 139.)

According to Blesser and Salter (2009, p 72) spaces and their relationship to sound have had a great impact on cultural beliefs and general interpretation of the human experience of life. For centuries the use of various spaces and sounds has had an effect on a variety of cultures. By using phenomenons such as acoustic reverberation, the first forms of entertainment were born, as suggested in the following quote: “the reverberating sound of a cavern then becomes the voice of the cave spirit” (Blesser, B. and Salter, L.R., 2009, p. 72). As well as many people have used it to manipulate their audience, from religious to political leaders, as seen in the following quote:

“When the aural experience of an acoustic space is sufficiently strong, its voice contributes, however slightly, to creating an altered state of consciousness in listeners, even in modern listeners. By extension, the aural architect who designs a space is also an aural manipulator—a modern-day version of an ancient shaman.“ -(Blesser, B. and Salter, L.R., 2009, p 73).

According to this previous quote, the relationships between sound and space can have an impact on our experience of certain moments. When studying sound’s behavior in any space one constant phenomenon is observed. As exemplified by Anderson, Agus, Chen, and Lui (2017, table 2), the relationship between acoustic reverberation and the clarity of sound is a game of balance. The more reverb you perceive the less the sound perception is going to be clear and defined. Having this relationship in mind, we cannot help but ponder how would this impact us as recording engineers, interested in recording in reverberant spaces? Based on Georgiev (2010), the recording philosophy of transparent recording revolves around a truthful and honest representation of the performance inside the space. It is explained how important it is to have a realistic representation of the performance and an improved listening experience, but that also any recording is a subjective imitation of reality. Its counter philosophy would be creative recording, where the engineer has full control to alter the sound until he carves his desired sonic reproduction. Taking this into consideration, we would want to know if clarity and the amount of its presence inside different recording spaces, can be used to improve transparent recording? Can certain behaviors of the apparition in clarity be traced in different spaces, and how can we use them to improve a transparent recording?

HYPOTHESIS: By utilizing modern means of acoustic measurements, we can trace the behavior of clarity inside a space. If these patterns of clarity presence are discovered, we can utilize the results to predetermine the positioning of microphones and performers for the purposes of clearer transparent recording.

This paper will first define the clarity indexes that we are about to follow to determine if our hypothesis is correct and then illustrate their various uses. Then we will talk about certain practitioners of both acoustics and transparent recording. This will be followed by an explanation of our methodology and approach for how we intend on exploring clarity, and then we will present two case studies of clarity exploration inside two different churches and the recordings undertaken inside them. We will compare the results in hopes to answer our question. A conclusion will be drawn by a brief summary of our observations and achievements while predicting where can this project lead us in our future works.

Contextualization:

The ISO-3382 (2009), (International Standardisation Organisation), has laid standard rules and suggestions to inform proper acoustic measurements. Amongst the many topics discussed, there are 2 which are crucial to our project, the Clarity Index, expressed as C-50 (clarity in speech intelligibility) and C-80 (clarity for music intelligibility), as well as the Definition Index, expressed as D-50, according to ISO (2009).

• “D 50 is defined as the ratio between the energy in the first 50 ms to the total energy of the impulse response.”

• “C 80 is the ratio between the energy of the first 80 ms to the late energy of the impulse response.” (Anderson, H., Agus, N., Chen, J.M. and Lui, S., 2017. p. 1080)

Fortunately enough, in this day and age computers can calculate these in a matter of seconds while gathering the required information, see Farina (2000, p.11). According to ISO (2009), C-80 is expressed in dB and is analyzed over each frequency of the spectrum. However, it is also averaged as a total for each measurement. The same goes for D-50 which is expressed in percentage.

What are the Clarity and Definition Indexes used for in acoustic measurements? They are actually pretty versatile tools. Obviously they are used to determine aspects of sonic perception and behaviors. As seen in Hoehne, R., Roy, A. and Schroth’s work (1995, p.5), these indexes can be used to judge whether the perception of sound in a specific space will be clear enough. Another example would be yet again judging on clarity of sonic perception, this time in playback, (Hoehne, R., Roy, A. and Schroth, G., 1995. p 4). As seen in Chesnokov and SooHoo’s work (1998, p.6), C-80 and D-50 can also be used in creating listening spaces with the example of domicile environments. However, Echenagucia, Sassone, Astolfi, Shtrepi and van der Harten’work (2014) focuses on the uses of C-80 in the designing stage of an auditorium. The tools are versatile and can be used creatively for different purposes. As mentioned, our hypothesis is that we can use these indexes to study the clarity’s behavior inside different spaces and try to draw conclusions about whether or not they can improve transparent recording approaches. However, there are counter-arguments against these standardized tools. From what we gather, in their nature, they are simplified and the world of acoustics depends on many other factors to be fully understood at this stage, as seen below. “Obviously, the very complex structure of the sound field in a room may not be described entirely by the subjective criteria reported here, as well as by the objective measures D50 and C80 computed by a relative simple algorithm. That's why corresponding investigations for other partial criteria of audio quality should be undertaken. So are, for instance, spatial impression, reverberance, timbre colorations etc. Finding a relative simple objective criterion for changes in timbre has been subject of further experiments by the authors continuing until now (synthetic soundfield). Provisional results have shown the subject to be very complex and multidimensional. Judgements are strongly influenced by choice of the music motif.” - Per (Hoehne, R., Roy, A. and Schroth, G., 1995, p.12)

Even with such tools, clarity in acoustics remains a partially subjective topic. Many other factors need to be considered for pure objectification in real-life applications. As stated above, further research into other factors for clarity might have to be undertaken in future works.

Transparency is the art of delivering the most objective recording possible. There is no specific rulebook to perform transparent recording as long as the core value of truthful recording, without external alterations is preserved. Some microphone techniques employed would be Decca Tree, ORTF and Blumeline. Let us exemplify a couple of cases where transparent recording is explored. Arguably the biggest influence on this paper would be Georgiev N. (2010). London-based engineer, he is specialized in location recording and has performed countless recordings inside reverberant spaces utilizing both the transparent and creative approaches. His work goes beyond, as he creates convolution reverbs out of the very unusual locations he has worked in, for companies such as Acoustica, which allowed him to lecture on the topic of convolution reverb and clarity at the Audio Engineering Society Conventions, based on AES (2017). Interviewing him, led us to many realizations about our methodology and focus points. He could not stress enough the importance of clarity, and how it is strongly dependant on the reverb in a space. We draw heavy inspirations from his approaches and detailed pre-production methods in transparent recording. Recording and mixing engineer Sylvia Massy, famous for her unusual approaches to sound treatment, maybe the core definition of an engineer utilizing the “creative approach”. Notably being famous for recording bands like Tool, where she would employ just about anything to make the sound exciting. She even went as far as writing a book named “Recording Unhinged”, discussing various approaches to recording inside the studio, as seen in Massy and Johnson (2016). On the other hand, Massy is often found recording in locations such as dungeons, abandoned nuclear stations and even tube stations. As seen in her publications online, she utilizes very little microphones and sometimes portable recorders, such as simple 4 track Tascams, as seen in Massy (2018). She may not be a purist in transparent recording however, she employs techniques very similar to transparent recordings and gives them a fair share of postproduction treatment. Massy is arguably a big influence on this project, even if her core values do not fully lie with the ones of transparency. Andrew Hunt is yet again another London-based engineer. He, similarly to Massy, recorded a big production inside a church. While recording piano and vocals for Jonathan Holder inside St. Giles in the Field Church, he employed a mixture of both approaches as he did a stereo transparent atmospheric recording while also employing direct microphone techniques, as seen in SourceDistributionTV (2014). While scoring Interstellar, renowned film composer Hans Zimmer recorded the Pipe Organ of London’s Temple Church (Nolan, 2014). From the brief section, we can conclude that the organ was recorded with ambient microphones, as was the entire space. Zimmer clearly pinpoints the fact that they were there for the reverb of the space. However, he notes that he relied heavily on the artistic soul of his performer. Which is something worth mentioning, as to us, transparent recording also has the purpose of highlighting the performer’s true and real capabilities, without hiding them behind expensive effect units. It is hard to consider that the sound that made it to the film was the untouched one from the recording, but it at least lies in its foundation. Church Studios, located in North London, is a Church refurbished as a fully functioning recording studio complex. It is interesting to note the existence of such a location, which aims to combine the elements of natural reverberant live rooms, such as churches and professional recording facilities (Sound on Sound, 2015). We notice that there is not a fine line between transparent and creative recording. Almost every single time, both transparent recording and creative recording are undertaken to ensure the product’s quality. For this project’s sake, we will attempt to conduct our recordings as transparently as possible.

Methodology:

The very first step is to take various impulse responses inside the spaces, which would gather the C-80 and D-50 data we need. Here lies the first challenge. There are various techniques to take impulse responses, but there is not one considered the ultimate one. In fact, each has its own pros and cons (Boren, B. and Roginska, A., 2011, p. 2-3). A good choice is to generate a loud sound that is as short as possible. Gunshots, small explosions or similar, are the traditional approach. However, previous experimentations show that every shot is slightly different and non-replicable, which also means less accurate results but also not equally present throughout the frequency spectrum (Boren, B. and Roginska, A., 2011, p. 2). Another factor is that it is hard to convince people responsible for churches to take gunshots inside the space. The alternative is to simply play a sweep across to the frequency spectrum and record it. Based on Farina’s research (2000, p. 11-12), we will be taking an impulse with a logarithmical sine sweep method, which Boren and Roginska (2011, p. 2-3) claim are preferred for projects of our nature. This method is often used to measure exactly what we are looking for, clarity and definition, but it will also cause distortion (Boren, B. and Roginska, A., 2011, p. 2-3). The following is something we need to acknowledge: To generate a sweep we need something to reproduce it, precisely a loudspeaker. Even if loudspeaker manufacturers publish their speaker’s flatness on the market, they do not show you the extensive distortion in the 2nd to 9th harmonic orders, according to Senior (2018, p8). A good way to take care of this problem is to use speakers that are designed with that in mind. As seen in Appendix C, figure 1, the ATC SCM20A are the speakers that create the least amount of distortion, but they are on the market for about 5000 British Pounds, which is not very accessible for this project. We have access to Genelec 8810A for this project, which we acknowledge is most certainly distorting in the higher harmonic orders, even if further tests should be undertaken for more precision. To come to the point, the results will not be 100% objective, because our loudspeaker will generate some distortions in the reproduction of the sweep. Besides being reproduced, the sweep needs to be captured. According to Farina (2000), we would ideally use an Omni microphone, to gather as much information as possible. We theorize that if we use 2 microphones for each position, being the Behringer ECM 8000 flat measurement microphone and the main microphone of our recording, the Aston Spirit, we will have more detailed results. Results will differ between the microphones, as their responses are different (see Appendix C, figure 2 and 3). The more data we have, the more accurate our interpretations will be. On the other hand, the placement of the loudspeaker should follow the positioning of the sound source in the recording. Georgiev (2010) shows the importance of good converters when attempting such experiments, in order to to have a more accurate result. While interviewing him, he mentioned that Prism Sounds have the most accurate converters in his experience. However, due to budgetary restrictions, we will be forced to use what we have, Personus converters, which are also dissent but probably far away as accurate as the Prism ones. It is very important in such experimentation to record at the highest sample rate possible, according to Georgiev (2010), 96 kHz in our case.

Once taken, the impulse responses will be analyzed using the power of digital computing. First, Room EQ Wizard will provide us with important information such as C-80 & D-50 responses, it will generate Waterfalls and Spectrograms to visually inform us as accurately as possible, about the presence of each frequency inside the space. We are looking for the highest results possible, with C-80 we expect to have overall results above 10 dB and with D-50 we are expecting above 85%. We will also be looking into problematic frequencies during the sweeps, according to our software of use, these are the areas in waterfall charts that have an extensive presence, above 135 dB. Based on the results we will try to analyze and determine the behavior of clarity inside both spaces. And based on that we will determine a transparent recording approach. Since we will be placing one microphone based on our result, we want to place other microphones around it, but without considering any measurements. The goal is to compare the clarity in the different microphones and judge whether there is a difference in clarity between the one position based on the measurements and the ones that are not.

Case Study I: Chatsworth Baptist Church

Measurement Approach:

The measurements were performed inside Chatsworth Baptist Church with the following equipment: Genelec 8810A Monitor for Reproduction, Behringer ECM 8000 Measurement Microphone, Aston Spirit Recording Microphone, Personus Audio-Box 1818 VSL Interface and Room EQ Wizard Measurement Software. Our approach consists of a few logistical and pragmatic considerations. First of all, the loudspeaker would always be positioned on the spot where the performers would be located. Second of all, the measurement microphone would be positioned where the recording microphone would be located during the performance. This already creates a certain amount of limitations. The seats in this specific church were fixed therefore, the piano could not possibly be moved too much around the space. This led to considering that the measurement location for the piano would be in the center, front of the church’s altar. The measurements would be performed from 3 different locations, with 2 different microphones (measurement and recording) both with an Omni polar pattern, but also at 2 different heights (ear-height and 2.5 meters above ground), see Figure 2. Therefore, in total, we made 12 different measurements for one speaker position.

Interpretation of Results:

The very first thing that needs to be noted is that the Aston microphone is advertised as not particularly flat, with a boost in the higher end of the frequency spectrum (Appendix C, figure 2), which is shown correctly in every single measurement. However, in every measurement, the Aston microphone provides greater C-80 and D-50 results than the ECM 8000 measurement microphone, which also has a much flatter frequency response (Appendix C, figure 3).

C-80 RESULT INTERPRETATION: Let us first talk about the most important results for this project, the Clarity Index for music C-80. Please find all results in Figure 3. First of all, let us observe the C-80 results with the Aston Microphone at ear level. We notice a decrease in clarity the further away we go from the source. At 2.5 meters: 17.66 dB, at 6 meters: 11.23 dB, at 7.5 meters: 9.16 dB. At height we notice the same decrease, with 2.5 meters: 13.99 dB, 6 meters: 11.11 dB, 7.5 meters 8.19 dB. The same phenomenon is noticed with the results of the measurement microphone. If we now compare the same microphone and distance but look at the height changes we also have solid results that show that ear height has more clarity than 2.5 meters height. We do notice that the 2.5-meter measurements show greater difference with 3.65dB difference with the Aston and 2.22 dB difference with the ECM 8000. At 6 meters with have a much smaller 0.12 dB and 0.27 dB difference respectfully. Interestingly enough, at 7.5 meters we have a greater difference than at 6 meters, with both microphones having 0.97dB difference exactly. We can conclude that when close to the sound source, the difference in height with our C-80 results is noticeable, however, there is no proportion between distance from the sound source and height change. In all of our measurements, ear height shows slightly better results than 2.5 meters in height. However, the human ear can start noticing a small change in clarity perception with at least 3 dB difference (Hoehne, R., Roy, A. and Schroth, G., 1995, p.1), which is the case at 2.5 meters distance only. To summarise the closer we were to the sound source, the greater the C-80 results we obtained. In every scenario, ear height proved slightly better but easily negligible results for clarity as well. The only difference is that there seems to be a proportionate decrease in clarity as we went further, which is not the case with the change of height in different distances.

D-50 RESULT INTERPRETATION:

With our results for the D-50 Index of Definition, we notice a very similar phenomenon to our C-80 results. As before, there is a decrease in definition the further we go away from our sound source. As seen in Figure 3, with the Aston at ear level we have 2.5 meter: 97.1% 6 meters: 89% 7.5 meters: 83 % while at 3 meters height we have 2.5 meters: 94% 6 meters: 88.9% 7.5 meters: 81.4%. This constant decrease is visible in our measurements of the different distances and heights with the ECM 8000 as well. Yet again, similarly to our C-80 results, the D-50 results show that ear level has greater definition results than 2.5 meters height. With the Aston at 2.5 meters showing 3.1% difference in definition in favor of ear hight. Similarly, it shows 0.1% at 6 meters and 2.4% difference at 7.5 meters. With the ECM 8000, we have a difference of 4.2% at 2.5 meters, 2.3% at 6 meters and 3.5% at 7.5 meters. Yet again we observe that there is no proportionality in the difference of definition, much like in the difference of clarity, when it comes to the same microphone at the same distance but with different heights. However, we do notice that no matter what, the slightly better but almost negligible results, end up at ear height. To summarise, our D-50 Index of Definition results prove to be very similar to our C-80 results. This is very logical considering the given definitions for both indexes, as seen before. The closer we are to the sound source the greater definition we obtain. Height also matters as ear level shows to be slightly more define than our other example at 2.5 meters above the ground.

WATERFALL RESULT INTERPRETATION:

The waterfall charts for all measurements can be found in Appendix A, figure 1. The waterfall chart is an all in one tool, that helps us better understand a space’s acoustical properties. The X-axis (horizontal) represents the frequency spectrum, which we would be set between 20 Hz and 20 000 Hz (generalized human ear’s audibility). The Y-axis (vertical) represents amplitude expressed in decibels (dB). Therefore, at a 2D representation, we can read how present each frequency is. In Room Eq Wizard, we have colors representing how strong the amplitude is. When purple, the area is not very present, the more light blue it gets to more perceivable it is. When in green our frequency is audible, when yellow it is very audible but when orange and red it means that there is an extensive amount of amplitude in that area. Waterfall charts are, therefore, a useful tool to visualize the presence of specific frequencies inside the desired space. However, waterfall charts have also a Z-Axis representing depth. It represents the decay of the space’s reverb throughout the frequency spectrum expressed in Milliseconds (Ms). This is helpful to gather information on the ambiance of the space. After reading through the charts generated from our measurements (Appendix A, Figure 1), whose positions are described in the table of results Figure 3, we can gather a handful of important intel on our space. As noted previously our ECM 8000 measurement microphone captures a rather flat response whereas our Aston Spirit recording microphone has a boost in the higher frequencies part of the spectrum. This is very visible in the measurements as well as in the waterfall charts when comparing the same measurements with the different microphones. Furthermore, we observe a few problematic areas in our space. Almost in every measurement, the space is really shy on 20 to 50 Hz presence. A lack of sub frequencies is noticeable. Other problematic areas are rather in the extensive presence of certain frequencies notably around, 123 Hz, 129 Hz, 140 Hz, 148 Hz, 162 Hz, 176 Hz, and 319 Hz, which remain a constant problematic area in all of our measurements. By changing the position and the microphones themselves, we observe that the areas shift slightly but revolve around those specific frequencies. We can also observe that the change in height has a greater difference when the microphone is positioned closer to the sound source. Yet again the closer we are to our sound source, the more defined and less problematic our results tend to be. It should also be noted that the closer we are to the sound source, the more concentrated the problematic areas become. We do not notice massive changes however, we notice that the further we go away from the sound source, the more the problematic areas spread across the spectrum.

Recording Decisions & Rational: MICROPHONE PLACEMENT:

From our measurements, we gathered enough information to come up with a couple of recording choices. Based on definition and clarity measurements, we conclude that the 2 best positions are indeed Position 2 and Position 4, both 2.5 meters away from the sound source at ear level (C-80= 17.66 dB and D-50= 97.1 %) and at 2.5 meters height (C-80= 13.99 dB D50= 94%). Furthermore, we notice that these positions are also performing a bit better than the others in the waterfall charts, with Position 4 being a bit flatter and a bit less problematic. On paper, both positions are good, with Position 2 being slightly more favorable with a 3.67dB difference, which is the margin for the first small changes in human perception of clarity, (Hoehne, R., Roy, A. and Schroth, G., 1995, p.1). The reason why we choose Position 4 is that we had a royal piano and 5 singers positioned around it. Geometrically speaking, height gave us more direct coverage of the performers, who had to be positioned around the piano. Otherwise, with Position 2, the piano would be masking the singers. However, if the singers were in front of the piano, we would not be respecting our measurement’s conditions. Therefore, we decided to go with Position 4, which was a better compromise. Furthermore, we added two more microphones besides the Aston, without considering any measurements. We wanted to compare the predetermined positioning with a more randomized positioning. Actually they were not so randomly positioned. We respected the proportions of the Decca Tree technique, in order to maintain a good phase relationship. The two Sontronics STC-2 was a great choice because of their flatness and their possibility to switch between Omni and cardioid polarity. As suggested by Georgiev (2010), one of transparent recording properties is having a stereophonic image. Even if the results are looking into the main microphone, the stereo pair is also beneficial to the recording. Of course, our main reason to have them is to compare them to the center microphone.

REPRODUCTION OBSERVATIONS

We recorded multiple takes, with different performance styles. We present to you, two of them. The reason why is because we want to underline the importance of the performance and its relation to clarity and definition. By listening to the recordings at 96 kHz in SAE’s Mastering Suite, we notice right away that the more contained performance has greater clarity because some parts are more contained dynamic-wise. In the second take, everything is pretty similar in terms of clarity, with the exception of the last third portion of the recording. In the second take, a very emotional moment occurred for our lead singer, which arguably lead to greater performance. However, this completely changed the clarity in the recording. It is observed that there is great distortion at certain moments. This is our first conclusion: in scientific experiments using sine sweeps, clarity might have a more predictable behavior however, in a real-life performance scenario, it is greatly dependant on its relationship with the performance. Another observation is that the piano player seems to be very aware of the spontaneous dynamic changes in the second take, and tries to control his own dynamics based on the other performers. Unfortunately, the piano seems to be a bit out of tune at certain points, which also results in some muddy moments during both recordings. Timber, tuning, and harmony in that case, also seems to be of value for the perception of clarity. Overall we are happy with the clarity in the recordings, even if some moments tend to be sonically crowded. On headphones (using Sonarworks Reference 4 correction) we notice that there are a lot of background noises in the recordings that are perceivable in the quieter moments, (in the first take we can hear birds singing outside the church, while in the second, we hear siren passing nearby). On headphones, we also realize how important it was to position the stereo microphones to have a more exciting recording and much-needed ambiance. Since the main microphone was positioned on a very clear spot we have a much less significant EDT, also seen in Anderson, Agus, Chen and Lui’s example (2017, table 2). Therefore, this relationship between ambiance and clarity explains why the center microphone has great clarity but lacks in ambiance. Which is the opposite of the stereo pair? Since the clarity in these spots was lower, the reverberation becomes higher. It is highly perceivable when isolating the center microphone from the sides, and AB testing between them, that there is a more significant amount of clarity in the microphone that was positioned after the measurements. We can conclude something quoted in our contextualization from Hoehne, R., Roy, A., and Schroth, G. (1995, p.12) and it follows that C-80 and D-50 are great tools, but clarity also depends on the sounds that we are recording and their full form. In this case, the arrangement is quite powerful at some points in the second recording and compromises the overall clarity. In order words, to have more control over the clarity, we need to explore other factors besides positioning. We can even go back to the quote at the very beginning of our paper. Vitruvius finishes by talking about the science of harmony. Which ends up being an important factor as we have seen. Both the dynamics of the performance and the harmony of the music did impact the sonic clarity. However, we see that the use of measurements can definitely help improve clarity.

Case Study II: St Anne’s Church

Measurement Approach:

The measurements in this location were performed with the exact same equipment as in the other space. This space had a very different architectural style and overall aesthetic. It is worth mentioning that St Anne’s church is roughly 300 years old and the acoustic properties were probably envisioned with a specific style of performance. The space is much more reverberant than Chatsworth Baptist Church. We considered that the stage would be the most pragmatic choice for our performers, which led to the first difference in approaches. We had a big stage and 20 singers that could easily fill it. This meant we could take measurements from multiple sound source locations. We took our measurements from 3 different speaker positions: front-centre of the stage, mid centre of the stage and on one side of the stage, as seen in figure 5. Here is where we could have improved and taken measurements from both sides, and regret not doing so. Also for a perfect result, we could have taken them from different depts of the stage, not only two. Unlike the Chatsworth measurements, we did not experiment with height, simply because the stage was already really high and if we wanted to capture the entire choir’s direct signal, we had to go as high as we could, please note that our stands could not reach higher than 2.5 meters, from the ground to the capsule. Therefore, we took measurements with 3 different speaker locations, for each we took a measurement from 2.5 meters, 6 meters and 7.5 meters away. Of course, each measurement was taken with the 2 microphones.

Interpretation of Results:

C-80 RESULT INTERPRETATION:

The results from the St Anne’s Church measurements can be found in Figures 7, 8 and 10. The first comparison we make is with the distance of the microphone from the sound source. With the sound source positioned in the front-centre of the stage, the Aston shows that the clearest spot is 2.5 meters away with 11.17 dB of overall clarity against 7.72 dB and 5.75 dB at respectively 6 and 7.5 meters away. Interestingly enough, the ECM 8000 shows that the clearest spot is actually the furthest with only 5.18 dB against 5.05 dB and 5.01 dB at 2.5 and 6 meters away respectfully. For the first time, the change in microphone proves an interesting difference, worthy of further research. However, let us be reminded that we start perceiving small clarity differences after 3 dB of change (Hoehne, R., Roy, A. and Schroth, G., 1995, p.1), which is hardly the case with the ECM 8000 results. They actually stay really consistent through the space. When the speaker is in the mid-center of the stage the results show the parallelism between the 2 microphones with both having the better results the closer they are to the source. The best result in C-80 measurements is also the Aston at 2.5 meters away from the sound source, located in the center of the stage, with 14.50 dB overall clarity, see Figure 8. Finally, with the speaker on the right side, we notice yet again a decrease in clarity the further away we go from the sound source as we would expect by now. It also shows very bad results with this speaker position, see Figure 8. Something, else worth mentioning is that when the sound source is in the mid-center of the stage we have greater results each time when compared with the other speaker positions, see Figure 8. This means that the choir should be positioned in the mid-center of the stage instead of the front. The best result is very firmly the 2.5 meter with the speaker in the middle of the stage.

D-50 RESULT INTERPRETATION: We notice a parallel-like behavior between the C-80 and the D-50 results. They do follow each other and are definitely related somehow and depend on one another. However, they are not proportionate. If you open the graphs provided by Room EQ Wizard, we see that they do not follow each other, but rather mimic each other. As seen in figure 9’s example, we notice that the curves follow each other, however, the D-50 curve (in orange) has a more sensitive response whereas, the C-80 (in green) has a more controlled sensitivity. It may be due to the fact that C-80is expressed in dB and D-50 in %. This is to say that the same conclusions can be drawn for the definition index as for the clarity index. The best result is the Aston at 2.5 meters from the speaker positioned in the mid-center of the stage, see Figure 10. Something else worth observing is that St Anne’s has generally much worse results in the C-80 and D-50 departments than Chatsworth (recently rebuilt), see Figures 8 and 10. Could it be because the technologies and intentions have shifted for the past 300 years in church architecture? Or simply because we did not take measurements at ear height in St Anne’s? Chatsworth actually felt like a safer and warmer space to record in from our experience as recording engineers. Further tests could potentially answer these questions.

WATERFALL RESULT INTERPRETATION:

The waterfall chart found in Appendix B, figure 1, shows us very general results on which we cannot base much. An overall observation shows that the low end in this space is less problematic than in the other, it is actually very well present. A couple of times we have a huge dip around 34 Hz, which is extremely narrow, it could potentially be a node of the room. The problematic frequencies are between 100 Hz and 700 Hz, which change slightly with a different position. The most problems are encountered when the speaker is located in the center-front of the stage. Interestingly enough, the flattest and less problematic results appear to be in when the speaker is on the side of the stage. When in the mid-center, where the results show the greater definition and clarity there are a lot of problematic frequencies, the further we are from the speaker the less problem. Through most measurements, we observe that the high end, above 1kHz, is nice and much flatter. Of course, the Aston always shows its high boost compared to the ECM 8000.

Recording Decisions & Rational:

MICROPHONE PLACEMENT:

This time around, we had to deal with a 20 person choir instead of a piano and a couple of singers. Based on our C-80 and D-50 results, we placed everyone as much as possible in the mid-center of the stage, where the speaker showed the greatest results. The lower voices were concentrated in the middle, while the higher voices were spread towards the sides. We explained the results to the conductor and she maneuvered her singers around. She explained that there were a call and response-like section, which forced us to keep specific singers together. We also tried to avoid putting people too much on the side, as the results were showing a lack of clarity. Nevertheless, we physically had to place certain people more to the side. Based on the results we had to place the center microphone where the measurements were better, at 2,5 meters away from the sound source and at 2.5 meters height. For this recording, we had the opportunity to record 3 Aston Spirits, which provides better criteria for comparison. The other two were configured similarly to the Decca Tree, just like the previous time. Yet again for the purpose of comparison, but proved useful yet again.

REPRODUCTION OBSERVATIONS:

When reproducing the recording on headphones, we noticed that the sound is clearer than what we expected, considering the nature of our performers. However, the dynamics can be quite drastic at times. With the center channel being the clearest, it has little ambiance and reverb, as expected. This is why the two other microphones are very useful to create width and add ambiance. However, they are distorting slightly, since they were not placed following measurements but rather according to the center microphone and the Decca Tree configuration, which led to lowering them down by 5 dB, in the stereo bounce. We notice many external noises in the quiet parts, which we could not fight against. On the studio monitors in SAE’s Mastering Suite, we notice the same points. The addition of the left and right channels is not as drastic, but it still improves the ambiance and the listening experience. Interestingly enough, we perceive less clarity in the recording when playing back on studio monitors than on headphones. The muddiness is also affecting some vocals which tend to fight amongst each other, which is far from being the case on the headphones. On the speakers, we also notice more distortion and more ambiance. We can for sure notice yet again that the relationship between ambiance and clarity lies in their rival nature. The more clarity we have, the less ambiance and vice-versa. Yet again the center channel has greater clarity in comparison with the added stereo pair, proving yet again that the measurements were beneficial to the choice of positioning. Certain moments of performance and the inaccuracy in timing also prove to be dreadful to clarity. We notice very similar traits to our previous case study, showing that the lack of harmony caused by bad timing affect clarity.

Conclusion:

Based on our measurements and recordings, we can conclude that the behavior of clarity can be predictable in large spaces, like churches. The closer we were to the sound source the greater results we obtained, each time. The same is applicable to the definition index. Of course, measurements will be more precise than your visual perception of the space and will help you better decide a recording approach. An interesting factor is finding the best position for the sound source, with which you can expect surprises, as we saw in our second case study. This is due to the uniqueness of each space and the relationship with the sound source. The practice of learning about the recording space is very beneficial. Some unexpected decisions were made based on it. Can transparent recording be improved by the study of a recording space’s clarity? Based on what we learned, it can. As we notice while playing back the recording, the microphone that was positioned based on the results had a perceivably greater clarity than the microphones that were positioned without the use of measurements. Since transparent recording is based on a truthful reproduction of reality, finding the clearest spot in the space for the sound source and the microphone is definitely helpful in achieving a good transparent recording. However, as we discovered through other people and our works, clarity is not affected only by positioning. Harmony, timbre, dynamics and musical motifs have an effect on clarity. Even if confined in limitations, this project has helped us understand more about clarity and its importance when recording in big spaces with transparent approaches. For our future work, we would like to explore other spaces with other proportions and ambiance, with examples of caves, tunnels and any other space with interesting acoustic behavior. We would love to expand into multichannel impulse response measurements and tackle the differences between the different arrays of recording. However, most importantly we would like to explore the relationship between clarity and timbre, harmony and performance itself, as suggested in the introduction and contextualization sections, but as well discovered through our work.

Reference List:

- Anderson, H., Agus, N., Chen, J.M. and Lui, S., 2017. Modeling the Proportion of Early and Late Energy in Two-Stage Reverberators. Journal of the Audio Engineering Society, 65(12), pp. 1017-1031. - Blesser, B. and Salter, L.R., 2009. Spaces speak, are you listening?: experiencing aural architecture. MIT press. - Boren, B. and Roginska, A., 2011, October. Multichannel impulse response measurement in matlab. In Audio Engineering Society Convention 131. Audio Engineering Society. - Chesnokov, A. and SooHoo, L., 1998, September. Influence of early to late energy ratios on subjective estimates of small room acoustics. In Audio Engineering Society Convention 105. Audio Engineering Society. - Echenagucia, T.M., Sassone, M., Astolfi, A., Shtrepi, L. and van der Harten, A., 2014. EDT, C80 and G driven auditorium design. Building Acoustics, 21(1), pp.43-54. - Farina, A., 2000, February. Simultaneous measurement of impulse response and distortion with a swept-sine technique. In Audio Engineering Society Convention 108. Audio Engineering Society. - Georgiev, N., 2010, November. Recording Philosophy, available at: http://georgievsound.com/ recording-philosophy/

- Hoehne, R., Roy, A. and Schroth, G., 1995, October. Comparing Investigations on Perceptibility of Sound Field Differences. In Audio Engineering Society Convention 99. Audio Engineering Society. - ISO, A., 2009. Measurement of room acoustic parameters part 1. ISO Std. - Massy, S. and Johnson, C., 2016. Recording Unhinged: Creative and Unconventional Music

- Massy, S., 2018. Recording Drums in a Decrepit Subway Station - with Sylvia Massy, available at: https://www.youtube.com/watch?v=dJ0XRFk0aQ0

- Nolan C., Nolan J. (2014). The Making of Interstellar, available at: https://www.youtube.com/ watch?v=L_8t2VlwK4w

- Pollio, V., 1960. Vitruvius. Dover Publications. - Senior, M., 2018. Mixing secrets for the small studio. Routledge. - Sound on Sound, 2015. SOS visit Paul Epworth at The Church Studios, available at: https:// www.youtube.com/watch?v=PCZ7cDR1L6c&t=277s&frags=pl%2Cwn

- SourceDistributionTV, 2014. Recording Studio in a Church with Andrew Hunt & Jonathon Holder, available at: https://www.youtube.com/watch?v=rJjhejzdXN8

Appendices: Digital appendices are available at: https://zzahariev27.wixsite.com/majorprojectappendix

The webpage also contains a logbook, the recordings, the separate tracks of each recording and a photo gallery.