Survey Results

European Cars Pack One was recorded as part of my Masters degree major project, the following text is a short extract from the report.

An online survey was conducted between the 22nd June and the 6th July 2012. It had earlier been piloted and refined with a group of 15 participants. The survey was targeted at people who use sound effects as part of their work or for fun. Of the 179 respondents that completed the survey, 77% gave their employment status as “Employed in Post Production” or “Freelance in Post-Production”. The survey was advertised on a number of industry websites, and the table below indicates where respondents heard of the survey.

Depending on the responses given, participants were asked up to 27 questions about their preferences concerning general sound effects and more specifically car sound effects. One of the key aims of the survey was to identify the type of car that consumers would be most interested in. Table 2 shows the responses of those who said they were likely to use car sound effects within the next year.

The clear choice of respondents was for Modern cars in common use in Europe and the US (see figure 1).

Of the respondents that were interested in cars in common use in Europe, 60% described themselves (or their employers) as likely to purchase a car sound effects library in the next year. Similarly, of those interested in cars commonly used in the US, 62% indicated a likelihood of purchase in the next year. Being UK based, the cost of recording cars in common use in the EU is significantly lower than capturing US cars. Additionally the survey showed that respondents based in Europe (those who selected Euros or GBP as their currency) did not have  an interest in cars popular in the US. In fact no European respondents selected this option at all.

The Renault Clio II was first introduced in 1998, and although the Clio III was introduced in 2006 (based on the Nissan B platform) the Clio II is still in production in 2012 under a number of guises (including Clio Campus). Approximately 8 Million Renault Clios were sold between 1990 and 2005 and the Clio appears in The Auto Insurance’s (2010) “Best Selling Cars Of All Time”. Furthermore, the Renault Clio regularly appears in annual lists of best-selling automobiles in Europe (EG. MSN Cars, 2012). While other cars would also be suitable to record, an experienced driver with a Renault Clio II has been identified. For this reason the following car would be recorded:

Model: Renault Clio II

Year: 2005

Engine: 1.2 L (1149 cc)

Type: 16-valve I4

Power: 75 hp

Top Speed: 170 km/h (106 mph)

Acceleration (0 – 100 Km/h) : 13.0 s

 

Sample Rate and Bit Depth

The survey also asked respondents about their preference of sample rate and bit-depth. The most popular responses were 24 Bit/ 48 Khz and 24-Bit/ 96Khz. However, a small amount of respondents preferred 16 Bit files. Some of those who preferred 16 Bit recordings noted that their libraries already contained many 16 Bit sound effects. While Protools and other DAWs can convert bit-depth of audio files (and most can handle files of different bit-depths in the same session), it was suggested by several respondents that 16 bit is still used in audio post-production making this their preferred format. As might be expected, many respondents identified the greater dynamic range available in 24-Bit recordings as an important benefit. As can be seen in Table 3, the most popular choice for sample rate was 96Khz. Many respondents noted that high sample rates (96Khz and above) introduced greater opportunities for pitch manipulation. However, for those that preferred 96Khz over higher sample rates, it is the trade off between the ability to pitch shift and disk space/ download time that was most important. Some respondents noted that they would like the option to select high and lower sample rates dependent on their specific needs. However, no specific application was identified as requiring higher sample rates than others. The necessity of high sample rates continues to be a controversial topic within the field of digital audio. Johnston (in Katz 2002) suggests that steep anti-alias filters at lower sample rates can result in a perceived loss of transient response. However he suggests that a sample rate of 50Khz is sufficient to avoid this effect at normal playback speeds.

Similarly Lavry (2004) argues that, for normal playback, a sample rate of 88.2 Khz would be unnecessarily high. In this case frequencies of up to 44.1Khz would still be accurately represented. Watkinson (2001) takes this further, suggesting that it is not possible for anybody to perceive a difference between a 96Khz and a 48Khz recording at normal play back speed. Watkinson suggests that any perceived improvement between 96Khz and 48Khz recordings is likely to be due to equally poor decimating filters employed in the conversion process. However in 96 Khz recordings the introduced noise is spread over twice the bandwidth. He argues that improving the converters used in the 48Khz recordings would mean the quality of the 96Khz audio would be matched.

Furthermore Lavry (2004) argues that sample rates as high as 192Khz introduce inaccuracies due to physical limitations of the circuits. It should also be noted that if a recording is made with a microphone which has a frequency response up to 40Khz then sample rates above 80Khz would introduce noise to the recordings. While a sample rate of 192 Khz can provide opportunity for extreme pitch shifting, in this specific case the drawbacks in terms of disk space, computational processing, possible loss of accuracy and introduction of distortion outweigh the benefits. The decision was been made to record at 24-Bit/ 96Khz.

 Planning What To Record

The survey responses shown in Table 4 (above) clearly illustrate that surround sound recordings are not a priority for those who were likely to buy a car sound effects pack in the next year. On the other hand, Multi-channel synchronous recordings were considered to be potentially useful or essential by almost 84% of this group. For this reason driving actions will be captured by several microphones from different positions.

Respondents were also asked about their feelings about the inclusion of several different types of sound. Table 5 records the responses of those likely to buy a car sample pack in the next 12 months.

Very few respondents considered any type of sample to be a “waste of disk space”. Demolition and Impacts were the types of sounds most likely to be considered unnecessary by participants, but even these sounds were thought to be desirable or essential by 68% and 76% of respondents in this group respectively. Respondents had mixed feelings about the inclusion of Composite recordings as part of the pack, with similar numbers of participants considering them to be “unnecessary” as those considering them “essential”.

The indication here is that the pack should include as wide a variety of sounds as possible. It is also essential to take into account financial and legal considerations as well as health & safety responsibilities to the driver and recordists when planning what actions to record.

It is noted that 91% of respondents felt that “Rolling Road” (Dynamometer) recordings were desirable or essential. During the initial stages of the research for this project, test recordings were made of a 280hp Fiat Coupe on a Dynamometer . A reading of 130dB SPL was taken approximately 0.3 metres away from the exhaust pipe. Nevertheless the “whine” of the Dynamometer was clearly audible, particularly during phases of deceleration even with careful microphone placement.

In retrospect, it is felt that this survey question was perhaps not as clear as it might have been. The key information absent here is why respondents felt this way. Deenan (2010) notes that Dyno recordings are primarily used to capture steady RPMs, as well as acceleration and deceleration ramps. Importantly it is possible to capture these sounds during road recording sessions rather than on a Dynamometer or Dynapack. There are significant challenges introduced when making road recordings (including microphone placement and surface audio), some recordists feel the results achieved are “organic” as opposed to the “clinical” recordings often achieved with a Dynapack/ Dynamometer. In order to capture drive- bys, aways, and approaches (considered desirable/ essential to 85% of the group), “real” road or track recordings would be a crucial part of this recording process. Although the ideal solution would be to plan both a full Dynapack and Road recording session, the budget does not allow for this luxury. Taking into account all of the above arguments, the plan is to record all car movements on a real road rather than an artificial environment.

You can download a selection of Pivot tables created from the survey data below:

RS Audio Survey Pivot Tables

 

 

Bibliography

Deenan, C. (2010) Charles Deenan Special: Car Recording Guide designingsound.org.uk [blog] 25 June. Available at : <http://designingsound.org/2010/02/charles-deenen-special- car-recording-guide/> [Accessed 10 August 2012]

Katz, B. (2002) Mastering Audio: The Art and the Science, Oxford: Focal Press, pp. 252 – 253

Lavry, D. (2004) Sampling Theory For Digital Audio [pdf] Available at : <http://lavryengineering.com/pdfs/lavry-sampling-theory.pdf> [Accessed 12 July 2012]

The Auto Insurance (2010) The Best Selling Cars of All Time [online] Available at : <http://www.theautoinsurance.com/best-selling-cars-of-all-time/> [Accessed 10 July 2012]

Watkinson, J. (2001) The Art of Digital Audio, 3rd Edition, Oxford: Focal Press, pp 729 – 730

0 comments on “Survey Results
1 Pings/Trackbacks for "Survey Results"
  1. [...] since then about techniques that worked, things that didn’t, and Richard even published the results from his survey about how car sound packs are used. It’s really awesome to see that he’s come to [...]

Leave a Reply

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>