External websites we've worked on: Dr Andy Moorhouse on piano ... Prof Trevor Cox
Before making recordings from products, it is important that they are used for a certain amount of time to ensure the parts have bedded in and so the noise is representative of their proper working conditions. For example, the washing machine should have been operated for at least 5 complete cycles; any load at rated capacity may be used for this operation. Similarly, the kettles were boiled a number of times before measurements. The products were used as a user would for the noise measurements.
The load consisted of cotton at a weight close to but in any case not more than the rated capacity. A 60 deg C cotton program was used without pre wash. If this program is not available the most effective program for white cotton according to the manufacturer’s instructions must be used. Special options selected by buttons (extra rinse etc) must be switched off. Detergent should be used. You should consider whether any auxiliary equipment such as electrical conduits, water piping etc. change the sound of the washing machine. For example, if the outlet is connected to the drain from a sink, excessive noise levels can radiate from the sink.
The noise from a kettle varies with the amount of water in the kettle. We recorded using one cup of water, the minimum fill, because it boiled quickest and was more energy efficient. A more comprehensive test might have used two different amounts of water and included both sounds in the jury testing. Before recording the kettle was filled with cold water and the element was used from cold. Lime scale may need to be considered as this affects noise levels. Manchester has very soft water an the kettles were brand new, so lime scale wasn't present.
We used a DAT recorder because they were available. This provides a high quality medium for recording. Alternatively, MP3 players which do not use compression (ones that don't use MP3, for example) might be used. Another approach is to record straight onto a computer through the sound card. The problem with this is that computers are quite noisy, and you do not want to pick up this noise in the measurement.
To faithfully recreate the sound made by a product, it is necessary to record the sound using a dummy head. A typical dummy head is shown to the right. This had microphones imbedded in the two ear canals, so the sound is recorded at the entrance to the ear canals. Importantly, the recorded sound then includes all the changes in the sound that occur due to the head, pinna and torso. This gives the listeners a better feeling of being within the space, and the sound from the product is more realistic. The only disadvantage of this system (called binaural recording), is that the system is quite expensive. The output from the two microphones in the ear canals are connected to a pre-amplifier and from there to the inputs of the recording device.
Alternatively, mono recordings can be made using a reasonable quality microphone. For example we recording with a single½ inch microphone connected to a pre-amplifier and the DAT player. However, the recordings are not so realistic. Our experience is that there is a reasonably good correlation between the sound quality results for monaural and binaural recordings. This might not be so true for products where the sound comes from every direction or is more immersive (for example a hairdryer or a motor car).
It is important that the sound recording settings remain the same for every product that is being tested. Changes in recording levels or room conditions will mean that the comparison between the different products is incorrect.
What to record
Ideally you should calibrate the recording system, so the sound can be reproduced at exactly the same level as the recording was made. To do this place a calibrator (an example is shown to the right) on the microphone and record the calibration tone. A calibrator is a device that produces a single frequency tone at a standard sound pressure level and are most often used to calibrate sound level meters. You can buy one from a sound level meter manufacturer. Check that the calibration tone does not overload the input levels to the recording device. Record 15s on the tape. If you don't do this, you will have to set the reproduction volume level of the products by ear, which isn't ideal, but can still result in meaningful results in the jury testing.
Obtain recordings that span the range of the effects you’re trying to measure. With modern computer software it is easy to edit the sounds afterwards. Sounds should be free of distractions (noise from instrumentation, unwanted squeaks and rattles from recording environment, talking/breathing/moving sounds from person making the recording, and other undesired characteristics that are not being investigated
Make sure the recording level is as high as possible to give the best signal to noise ratio, but not so high as to cause distortion. Recording devices have VU meters on them which show the level of the incoming signal. Make sure the readings are not too low or high. Your recording device manual will tell you what is an appropriate level. Your device will have a variable gain on the input so you can set the recording level to aim for. It is important that this gain setting is set to be the same for every product that you measure. This has to be true so the relative volume levels of the products are recorded correctly - you don't want the loudest kettle to suddenly become the quietest one.