Home Page Next Previous
This paper contains the accumulated experience of a number of people, including myself, in successful recording of bell sounds, and in conversion to a digital format. Digital formatting allows for storage of bell recordings on a computer, transmission by email, and publication on the world wide web. Some parts of this paper are specific to digitized recordings, but much is applicable to recordings kept in traditional analogue formats.
Since the paper was first published, a new section has been added on use of professional microphones, plus some notes on improving the quality of mp3-encoded recordings.
Andrew Wilby originally suggested I wrote this paper. It includes extracts from messages originally published on the Bell Historian’s chat list. The following people have helped with ideas and experience: Andrew Aspland, Alan Birney, George Dawson, Stephen Ivin, Dave Kelly, Nigel Taylor and Michael Wilby. Shahram Taherzadeh of the Open University helped me with microphone calibration. Errors and unsubstantiated opinions are my own.
There was a useful article on this subject in the Ringing World of 24th April 1987, pages 379 and 380.
This paper covers the following subjects:
Similar issues on the Macintosh should also be covered but unfortunately my knowledge of Macs is sketchy.
Details are given on another page on how to do an accurate frequency calibration of your recorder or PC.
Bells have been recorded on many different devices. Examples using analogue media include reel-to-reel tape recorders, cassette recorders, video-camcorders and even dictating machines! Examples using digital media include recording direct to PC or laptop, mini-disc recorders and mp3 recorders. The main issues in selection of a recording device are:
Microphone quality is obviously important. Many devices have built-in microphones, and these should be regarded with suspicion (though Dave Kelly and myself have both had good results from the built-in microphones in video camcorders). External microphones are available very cheaply (a few pounds) that give excellent results, provided the recording device has a microphone input socket. Wide frequency response is important: 50 Hz to 15 or 20 kHz will be needed to capture the full range of partial frequencies. Condenser (sometimes called electret) microphones are said to give better response to the brief, high intensity, high frequency sounds emitted by bells than dynamic (sometimes called moving coil) microphones. For details (and the problems of powering microphones) see the section below on using professional microphones.
Though cheap microphones often work well, cheap and small loudspeakers are usually very poor, with bad bass response and fake resonances. You need to play recordings back through hi-fi equipment (or at least PC speakers that cost more than £15!) to judge their quality.
Frequency stability of the recording device is very important, especially if the recording is to be used for tonal analysis or pitch determination. Frequency stability is not an issue with digital recorders, and nor does it seem to be a problem with camcorders. Cassette recorders, unless they are very good quality, are poor in this regard. When pitching bells, it is possible to get around frequency stability problems with the use of tuning forks or reference handbells, but this is somewhat inconvenient.
Recording devices with automatic level control should be avoided like the plague. Because bells should be recorded in a quiet environment, and are very loud, recordings taken with automatic level control usually consist of loud background hiss interrupted by the fainter sound of the bell or bells. Digital recording devices and camcorders seem not to have auto-level control.
On the other hand, recording devices with manual level control are very useful. Bells are very loud, and can easily overload the recording device. Manual level control can be used to turn down the input volume and produce acceptable recordings. Without this, the only option is to move the microphone further from the bells, which is often not desirable or possible. The section below on recording techniques has more advice on this.
Steve Ivin and George Dawson report considerable success with minidisc recorders, though Steve says his Sharp minidisk suffers disc errors when used in the bell-chamber. Andrew Aspland has taken recordings successfully on an mp3 recorder. Good results can be had using either an analogue camcorders or by recording direct to laptop. The only recording I know of taken with a dictating machine was terrible.
Bells are very loud indeed, as we all know. I have found that a single clapper blow recorded close by the bell can easily overload the various recorders and microphones I use. There are two ways to get around this. First, distance always helps - recording from another floor, further from the tower etc. Learn the characteristics of your recorder and how close you can get without overload. The second cure is to always do a level check before taking a final recording, and either move further away or turn down the record level to get satisfactory results.
Just because the recording is not audibly distorted does not mean that you are free of overload - looking at the file with a sound-file editor will tell you if clipping is occurring. Aim for no more than 80 or 90% of full scale to be sure. If it is not possible to view the waveform in this way and your recording device has a meter, do not allow the record level to go above 5 or 10 dB below the maximum. The brief but high intensity partials produced by the clapper blow die away before the meter can register them. Aim low - within reason, loudness can be boosted with a sound file editor but there is no way to remove overload.
I am now using a professional microphone, able to cope with high sound pressure levels (SPL), for close-up recordings with excellent results. Both Andrew Aspland and I have had good results from high quality microphones suspended directly above a peal of bells.
As a rule of thumb, recordings I have taken of average strength clapper blows on a camcorder 15 or 20 feet from the bell have proved to have clipping.The plot on the left (from a sound editor) shows an overloaded recording.
Ambient noise and recording location are important. The ability of recorders to pick up blackbirds, traffic, wind and bellropes is remarkable. Wind can to some extent be cured with a hood on the microphone. The background noise needs to be minimal for a good recording. Watch for multiple sound paths from the bells to the microphone. Examples include: echoes off surrounding buildings, closeness to a bell-opening, or to a hole in a floor giving undue prominence to certain bells, etc. Locations inside the tower are often satisfactory, especially if it is possible to go above the bells with an intervening floor or two to cut the volume down. On floors below the bells, rope or mechanical noise often spoils the recording. If it is sheltered from the wind, the roof of the tower may be a good location. The church roof is often too close to the bell openings on one side of the tower. Some of the very best recordings are taken in the bell chamber, but need a professional microphone capable of handling the sound intensity. Careful positioning of the microphone is needed to avoid giving prominence to individual bells.
Intervening floors or walls give selective attenuation of frequencies. This does not matter if the purpose of the recording is just to present the overall impression of the bells as heard by the man in the street, or to preserve a special piece of ringing. On the other hand, this selective attenuation matters very much if comparing the timbre of bells with those in other towers.
If recording individual bells for the purpose of frequency analysis, do not record them while they are swinging. The Doppler shift due to the moving bell will introduce considerable confusion into the analysis. Either ring the bell onto the balance, or ring it with the clapper by hand while it is hanging stationary. The latter gives the most accurate results.
This subject can be divided into two parts; the recording parameters used (sampling rate, sample size and number of channels) and the file or compression format used.
The greater the sample rate, sample size and number of channels the bigger the resulting sound file. Therefore, there are trade-offs to be considered. It is almost always better to reduce the file size through compression rather than by compromising the original recording format.
A theorem known as the Nyquist Theorem, grossly summarized, says that the sample rate should be double the maximum frequency present in the sound. In practice I find with recordings you need to go higher than that. For the best results, sample at 44,100 samples per second. Recording at 22,050 runs the risk of losing frequencies somewhat below 10 kHz but may be acceptable if the original recording is bandwidth limited. Do not digitise at 48,000 as some Windows drivers have bugs at this sampling rate.
As regards sample size, 8 bit data gives disappointing results, 32 bit data is complete overkill and is not universally supported. Use 16 bit. Except for recordings in exceptional circumstances (e.g. recording a swinging bell close up) I do not believe stereo adds any value to recordings of bells. Save the disc space and use mono!
There are three file / compression formats regularly used for bell and other recordings on PCs; uncompressed waveform files, compressed waveform files (both with file extension .wav), and mp3 files (with extension .mp3). Mp3 format is often referred to as MPEG layer 3. Other formats sometimes used are aiff, .au and .ra files. These latter formats are not as common and not all web users will have the software to receive and play them, though the Real Audio player needed to play .ra files is becoming ubiquitous. If uploaded to websites, .wav and .mp3 files do not need any special software on the server. Real Audio files require special server software at the ISP or web host to send the files to users. The remainder of this section considers only .wav and .mp3 files.
Files recorded in raw .wav format (often called PCM or Windows PCM) at 44,100 samples per second, 16 bit, mono use over 5 Mbyte of disc per minute of sound. These files are uncompressed and therefore reproduction will be exact – at the expense of enormous file sizes.
Wav files can be compressed. The compression algorithm called Microsoft ADPCM is installed on all PCs running Windows 95 or later and provides a 4 to 1 reduction in file size over PCM. Disc usage is therefore about 1.3 Mbyte per minute of sound. The compression process is lossless, and in trials I could detect no audible difference between PCM and ADPCM format files.
Mp3 files are very compressed using a ‘lossy’ compression algorithm. The format I use (44,100 16 bit mono compressed to 56 k bits/sec) compresses files by a factor of 12 to 1 over raw .wav files, so that 1 minute of sound occupies about 420 kByte of disc. If the original recording is high quality the mp3 format gives reasonable reproduction. However, if the original recording is distorted due to microphone or recorder overload, or has a ‘wash’ of background bell harmonics, mp3 compression will give very poor results. Older PCs may not have the right drivers to play mp3 files, though these can be downloaded from the web (see the next section). The drivers come as standard with modern PCs.
In recent experiments with mp3 format recordings, I have found a way to improve their quality by reducing the bandwidth of the recording before encoding it into mp3. I believe the poor quality of some mp3 recordings of bells results from high frequency artifacts due either to distortion and overload, or lots of high frequency partials in the bell sound. The mp3 encoder is reproducing these high frequencies at the expense of lower frequencies. I have found that considerable improvement results from using a sound file editor such as Cool Edit to reduce the high-frequency component of the original waveform. In experiments, I used the 'FFT filter' facility of Cool Edit to reduce frequencies above 12 kHz by 15dB and was pleased with results. I am interested to hear of other's success with this approach.
As an interesting aside, the GSM compression algorithm used on mobile phones is also supported on PCs. Trials suggest that GSM compression gives reasonable reproduction. Dave Kelly reports that he has successfully analysed bell sounds transmitted by mobile phone, though frequencies below 300 Hz and above 3 kHz are considerably attenuated.
As a rule of thumb for file sizes, a 100 kByte file will take 15 or 20 seconds to download at 56 k bits/second. For recordings in 44,100 16 bit mono, this file size will give:
Usually the whole file must be downloaded before the sound will begin to play. For Real Audio files, this is not the case, the file is streamed and will begin to play after a short delay. Mp3 software now also provides this facility.
To record or digitise sounds using a PC, software to record and edit sounds is needed, plus either a microphone or a cable to connect the PC to the recording device. A sound-card is needed in the PC, which these days may be integrated into the motherboard.
PCs or soundcards have up to four sockets selected from the following list:
These may be identified by words or symbols. The line-in socket should be used to connect to an external recorder. The microphone should ordinarily be plugged into the microphone socket – condenser microphones intended for use with PCs pick up the power they need from this socket. Steve Ivin suggests plugging a microphone into the line-in socket if the sound being recorded is very loud. I do not have any experience of doing this, and use instead an attenuator - see below in the section on professional microphones.
Cables connecting the PC to an external recorder can be bought in any high-street electronics retailer. A 3.5mm stereo jack is needed at the PC end. The connector at the recorder end is likely to be either another 3.5mm jack or a pair of phono plugs.
The simplest and cheapest software for recording and digitising on PCs is Windows Sound Recorder. If you do not have an icon for this, it can be activated by running the program ‘sndrec32.exe’ from the Start menu. Sound recorder is very simple – it does not have any effective edit facilities – but it can be used to convert sound files between the various formats described above. However, it is not obvious how to change the recording and compression format.
This can be done in two ways. Either follow the menu tree File : Properties : Format Conversion : Convert Now, or reach the same dialog via File : Save As. In either case, select the appropriate recording format as explained above. (If no mp3 formats are listed, you will need to install an mp3 codec on your PC. Search the web for ‘fraunhofer mp3 codec’). Once the format has been selected in this way the sound should be digitized in the correct format. I note with interest that Sound Recorder on my laptop does not support the full range of mp3 formats. The picture on the left shows the dialog in Windows Sound Recorder allowing the file format to be changed.
To provide more flexible recording and to allow editing of sound files, a sound-file editor is required. Sometimes one is supplied with the sound card, but more powerful packages are available as shareware. My personal recommendation is Cool Edit, which I have found very useful. It is shareware, but you can download a trial version and use it for 30 days (with restrictions) before you have to purchase it. It is available from www.syntrillium.com. Its most useful facilities are:
I use this package extensively to pick individual bells from a recording, edit out initial silence, trim recordings to length, and remove noise. Other packages are also available.
PCs have a recording level control useful both for recording direct from microphone and when digitizing previously recorded sounds. This can be accessed from within Cool Edit by following the menu tree Options : Windows Mixer, or with any package by double clicking the loudspeaker icon on the Start bar. There are different versions of this program, but typically what appears first is the playback mixer. To access record levels follow Options : Properties and click the ‘Recording’ radio button. Then use either the microphone or line-in slider depending on which input socket you are using. The settings appear to be saved automatically when they are changed. The picture on the left shows the record control on my laptop
Some PCs also have the facility to include extra amplification of a microphone signal. If this feature is available, it is usually to be found by clicking a button marked 'Advanced' on the Windows Mixer, accessed as above. On some machines, including my laptop, it is necessary to select 'Microphone' as an output device (!) before this button appears. The extra amplification can be useful. However, you should be warned that on all PCs I have tested, the amplification also considerably increases the input noise, to the extent that recordings can be very poor quality. The input noise in a laptop can depend on whether it is powered from battery or mains. My laptop is much quieter when mains powered. (On the other hand, I get mains hum when using it on mains power with my standard PC microphone.)
If you make a lot of recordings, you may consider buying a professional microphone, for a number of reasons:
When selecting a professional microphone, there are a number of issues and pitfalls to be aware of. It has taken me a considerable amount of investigation to understand this subject; here, for the benefit of others, is what I have learned. The issues are:
Professional microphones come in two basic types: condenser (also known as electret), and dynamic (also known as moving coil). Dynamic microphones may be less expensive, but produce low output volume, and may cut back the high frequencies. Condenser microphones on the other hand usually require some form of power. Dynamic microphones may also need power if they include an amplifier in the microphone.
Professional microphones are powered by 'phantom power', nominally 48 volts, supplied by the amplifier to which they are connected. PC soundcards usually provide microphone power (because most if not all PC microphones are condenser), but only 5 volts nominal which is quite inadequate. Other consumer recorders may not provide power. So, unless you are prepared to invest in a suitable pre-amplifier, it is necessary either a) to be certain that the microphone you are considering does not require phantom power, or b) to purchase a microphone which has provision for a battery. High quality condenser microphones with provision for an internal battery are becoming increasingly common.
PC soundcards (and I suspect other domestic recorders) often have quite insensitive microphone inputs; i.e. they require a decent voltage to get a recording of reasonable loudness. Dynamic microphones often have quite low level output. Turning up the gain in the sound card or recorder can compensate in part for this. However, this also significantly increases the noise, which may negate the reason the microphone was purchased. I have no personal experience of using dynamic microphones, but read that they often produce poor results for this reason. As a guide, a microphone with a sensitivity of 5 millivolts per Pascal (often described as -45 dB) will produce good results when used with a PC soundcard. Dynamic microphones with lower sensitivities (e.g. 0.2 millivolts per Pascal or -75 dB) will be too quiet.
On the other hand, some condenser microphones may produce too much signal for the soundcard or recorder, especially if the sound being recorded is very loud (e.g. a peal of bells, close up). In this case it may be necessary to drop the input voltage via an attenuator or pad to avoid distortion. I can provide details of this for those who are interested.
The 'impedance' of a microphone describes its ability to drive a load (such as the microphone input of a soundcard). Microphones come in two basic types: high impedance, and low impedance (sometimes characterised as 10 k-ohm and 600 ohm respectively). Some microphones can be switched between the two. Only low impedance microphones will work properly with sound-cards or similar devices.
Professional microphones usually drive a 'balanced' signal onto the cable. This gives much better noise resistance than the unbalanced interface on the cheap microphones which come with PCs. A low-impedance microphone with a balanced output can drive many tens of metres of cable successfully. This allows, for instance, a microphone to be hung above a peal of bells and the recorder to be located in the ringing room or suchlike convenient place. However, it is necessary to convert the balanced signals from the microphone to unbalanced signals for input to the soundcard or recorder. This can usually be done via wiring in the connector, but this may reduce the noise immunity, frequency response and permissible cable length.
Most, if not all, professional microphones use an XLR connector. This is a three-pin connector with two balanced signals and ground. The microphone input on soundcards and many consumer recorders is a 3.5mm jack with signal, ground and sometimes a 5v supply for a condenser microphone. Electronics and hi-fi shops will sell a cable to connect the two. The correct wiring of the cable is:
A cable wired in this way is not balanced and will not provide much noise immunity. A cable wired in this way should be kept quite short.
With the help of the Open University I have recently done calibration experiments on a number of microphones including an AKG C1000S professional condenser microphone I recently purchased, my trusty Blaupunkt CC824 video-camera, and a cheap PC condenser mike. The AKG was tested both via a direct (unbalanced) connection and using a home-built balanced-to-unbalanced converter incorporating a high quality microphone transformer and a 20:1 switchable attenuator.
The AKG microphone produced excellent results. It gave distortion-free recordings when hung a few feet above a heavy eight rung full circle, though the attenuator was needed to drop the signal to a level compatible with my laptop microphone input. The microphone had no problems driving 20m of balanced microphone cable. The frequency response of the AKG was very good, being substantially flat between 95 Hz and 6.3 kHz when used via the transformer. Connected directly the lower limit rose to 120 Hz. The microphone response is specified over a much greater frequency range (50Hz to 20kHz) but I was not able to measure to these limits. Used via the transformer, there was no significant mains hum. Connected direct, the hum reached 15dB above the background noise, not catastrophic but worth noting.
The video camera, which I have used for a number of years, overloads on close recordings but produces good results on quieter recordings. Its frequency stability is very good indeed and requires no correction. Its frequency response is flat from high frequencies (at least 6.3 kHz) down to 1 Kz. Between 400 Hz and 1kHz the video camera has about 5 dB of uplift. Below 400 Hz the response rolls off, being 13 dB down at 100 Hz.
The PC microphone is also unusable for close recordings due to overload, but works well for quieter sounds. It suffers very severe mains hum when used with a mains-powered laptop. On batteries, its mains hum is about 25 dB above the background noise. Its frequency response rolls off uniformly from high to low frequencies, being 17dB down at 100 Hz.
Last updated October 30, 2002. Site created by Bill Hibbert, Great Bookham, Surrey