The second installment in the thriller… Note that this section has not yet been revised and was written before a year of using a Marantz compact flash recorder.
Even if your recorder is excellent, your recordings will be bad if you have a low quality microphone. There are several different types of microphones, and which one is best depends on your recording situation, your recorder and how much you can spend.
Microphones can be mono or stereo. Mono microphones record a single channel, whereas stereo microphones record two “streams” or channels from different directions. Stereo microphones will give more sound depth and are useful when you have several speakers. A single channel is fine for most types of language recording.
We can also classify microphones according to the technology they use to relay the sound to the recorder. Condenser microphones are one type. They have better fidelity to the input sound but they require their own power supply, they are quite fragile and can be susceptible to changes in humidity. Dynamic microphones produce a worse quality recording but they are much more durable and they do not require their own power source.
Another distinction is between low-impedance and high-impedance microphones. Low-impedance microphones more accurately reproduce the sound being recorded, but they are more expensive and more fragile.
The other most common feature of microphones with the shape of the target recording area. A shotgun microphone is one that only records a narrow area of sound directly in front of the microphone. They are highly directional and are better in areas of ambient noise. Note, however, that they don’t record from any further away. They also need to be placed more carefully. A cardioid microphone records a ‘heart-shaped’ area. It is maximally sensitive in front of the microphone head but also records at an angle, so they are more forgiving if the speaker is not directly in front of the microphone. Cardioid microphones come in different degrees of directionality, and some have the options of recording at different angles. The third common shape of recording area is ‘omnidirectional’ – that is, the microphone records from all angles.
“Business” microphones tend of be omnidirectional. These are excellent if you have a group of people – you can put the microphone in the middle of the group. But they aren’t so useful if you are mostly recording a single speaker, as they will pick up a lot of background noise.
<<picture of recording area.>>
Some microphones have variable settings for recording areas; e.g. a narrow setting and a wide setting. These are very useful if you record both single speakers and groups of people. They are also useful if you wish to use your recording equipment for different types of recording, such as music as well as speech..
Lapel and head-mounted microphones are also used by some people. These are good if you are just recording one speaker. They produce very good recordings of the person weaning them and eliminate as much background noise as possible. The lip (head-mounted) microphones eliminate noise which is coming from directions other than the speaker’s mouth (the technology is used in pilots’ microphones, for example, where there is a large amount of external noise.
I don’t use these type of microphones because they are more difficult to forget about and they counteract my aim of producing a non-threatening recording environment (i.e. one that is a general conversation that happens to have a tape recorder running). Also, I never know how many people are going to turn up to the session and how many people are going to tell stories. Furthermore, it can be helpful to use your recorder as a note taker, in which case you will want to record the questions you are asking as well as the answer given. But they are highly advantageous for recording phonetic data.
Lapel microphones are usually omnidirectional and thus can pick up a lot of background noise, although in their favour they are less obtrusive than head-set microphones. They can also (in theory) be put on a stand in the middle of a group. Note, however, that some of the better lapel microphones are designed to be placed on the speaker’s chest cavity, and in fact they pick up the bottom frequencies from the chest cavity rather than from the oral speech signal, so using them on a stand significantly changes the frequency responses. They are easier to forget about than lip microphones or head-mounted microphones, but wearing them as lapel microphones presupposes that the speaker is wearing clothes that the microphone can be clipped to (you do not want the pop shield of the microphone touching anything). In some cultures it might be inappropriate for a male researcher to touch a female consultant (or vice versa) to clip on the microphone. In that case either use a microphone with a stand or demonstrate on yourself and have the consultant clip the microphone on themselves.
Many of the higher end microphones require pre-amps and their own power supply. As always, there is a trade-off between quality and portability. You will get the best recordings in a sound booth or radio station, and in some cases you might be able to use this for part of the time. Many Australian Aboriginal communities have a BRACS studio, although the amount of equipment and the quietness of the air conditioning vary greatly between communities.
2.3 Types of audio recorders
2.3.1 Analogue cassette recorders
Analogue recorders are fast going out of fashion in many places. It’s increasingly difficult to buy good quality ones, the best quality recorders are more expensive than digital recorders which make better recordings and are easier to get data off. In their favour, they are easy to transcribe from, and in many parts of the world a lot of people still have tape players, so your consultants may find them easier to play. Also they are easier to get data off if you have a partial break-down. You can splice tapes back together, for example, but if a CD or MD breaks you will lose all of the data on it. The media are cheap, which is good, and widely available (although increasingly less so).
Against them is the problem that you have to digitise the recordings if you want to do much with them, such as making spectrograms from the recordings, editing them by playing segments in real time to make talking books or dictionaries, copying them onto CD, etc. Tapes can also become demagnetised over time, they need to be kept out of the heat, and they can warp. Power fluctuations influence the speed of the tape and the pitch of the recording. So, I strongly recommend starting with a digital recorder in the first place.
2.3.2 MP3 and MP4 recorders, and IPODs
These recorders come in various brands and varieties. Most of them have built-in memory that allows you to record for a certain period of time, then you have to download the recording to your computer. They are reasonably cheap, highly portable, have few moving parts (which means they are robust) and often have pretty good battery life.
On the other hand, the quality of recording is not good enough to do phonetic analysis, with very low sampling rates (that’s how you can fit a few hours onto a 256Mb chip). Also, mp3 is a compressed format, which makes it hard to use for spectrographic analysis (and how knows what other information is being lost on mp3 recordings that linguists in future might find useful)? The algorithm that computes the compression is proprietary, and so if the algorithm is changed it means it will be impossible for us to read the format. The built-in microphones that these devices come with are usually very bad, too. The same is true, at the moment, of IPODs. Some types of IPOD have the capacity to plug in an external microphone to make recordings, but at present you cannot change the sampling rate, which is set at 8,000 Hz.
Until the possibility for better quality recording becomes available, I can’t recommend these recorders except as a last resort backup. E.g. if you have an IPOD and are taking it to the field to listen to music know that you might be able to use it as a backup in an emergency.
2.3.3 DAT recorders
DAT recorders have been a favourite with musicians and musicologists for a while. They make excellent recordings with high sampling rates. There are small, portable units now available which cost about the same as compact flash recorders, and are cheaper than CD-recorders.
DAT recorder have drawbacks for linguistic field work, though. The equipment is expensive and can seize up and break down with sudden changes in humidity. The media is expensive and tapes are only available in specialist shops (although the media is reusable so if you can transfer your recordings to another format regularly you can reuse them).
Another disadvantage of DAT recorders is that to transfer your recordings to your hard drive you basically have to redigitise the tape. That is, transfer has to occur in real time (by playing it back); many sound cards can’t cope with digital input from DAT recorders and they treat the signal as analogue and re-convert it. This still produces a higher quality recording than a cheap analogue tape in the first place, though.
We don’t know much about the longevity of DAT tapes, and the format is already being superseded by compact flash recorders. They are not a good archival format either. The recorders sometimes require mics with pre-amps, which means more equipment to carry around and more batteries to charge. DAT recorders don’t have a great battery life. They work best with a mains supply.
Therefore if you are thinking of buying a DAT recorder, I would recommend a compact flash recorder instead (see Section 2.4.7); they are approximately the same price.
2.3.4 CD recorders
There are recorders which can record directly onto CD. They again are able to record at high sampling rates (44,100 Hz). Because they burn directly to CD you do not have to redigitise, and it’s easy to make copies of your recordings. You can add track marks to the recording. You can record from several different inputs (that is, you can have several different microphones) and the channels can be merged or split
Again, however, there are cons. The machines are somewhat fragile and not very portable. They aren’t run off batteries, so you will need a good power supply. They are also rather expensive, over US$1,000 (compared to about $600 for a portable DAT or compact flash recorder, and $250 for a good minidisc recorder). They are somewhat unintuitive to use, and the disks need to be finalised before they can be ejected from the machine. This takes about two minutes, which means that you cannot quickly switch disks to continue recording. The model we used in class in 2005 needed five key pressed before recording started.
2.3.5 Recording directly to computer
Another possibility is to make recordings directly onto your laptop computer. You can pug an external microphone into the computer and record directly to the sound card. Your recordings are saved digitally. To do this you need a sound card and the software to make the recordings. You should also have a lot of spare room on your hard disk drive and a CD (or ideally a DVD) burner to make backup copies.
This is another type of recording setup that is probably all right in an emergency or as a backup but is not workable for everyday recording without huge expense. There are problems with the quality of the recordings made; good sound cards are very expensive and with the cheaper models there are issues with noise from the processor, the hard drive turning, and distortion from the sound card itself. It’s intensive on the battery. Good software for audio recording and editing is also expensive. Windows is not designed for this sort of thing and you need to shut down a lot of processes to maximise the amount of memory dedicated to the recording. It takes huge amounts of processing power to do properly and requires vast amounts of spare disk spare if you are going to make this your regular recorder. Also, of course, all the problems that were mentioned above in relation to laptops also apply here; screens can be hard to read in bright sunlight, they are fragile, you can’t sit in the dirt with them.
Having said all this, there are professional musicians who have laptops with dedicated recording setups. There are also pre-digitisers (or external sound cards) which can be used to modify an existing computer system.
There is a lot of conflicting information about minidisc recorders and their applicability to linguistic fieldwork.
Minidisc recorders are handy – they are very small (therefore highly portable), and they have excellent battery life (and most of them take AA batteries). The media are quite cheap if bought in bulk (as low as US$1.50 per disc) in some parts of the world and are rerecordable. Some of the more recent models can record in linear pcm format (that is, uncompressed audio) which removes one of their earlier main objections – that is, that the compression and proprietary format of the audio recordings are a bad idea. The files can be directly uploaded to a computer and saved in .wav format.
Despite this, they appear to be despised by a large portion of the fieldwork community. The older models require redigitisation of the disc to get them onto a computer and into a format where the sound file can be manipulated (although the resulting recording is probably still better than an analogue recording from a tape recorder). The compression algorithms are proprietary so we don’t know exactly what they do to the signal. We have no data about the longevity of the discs and the back-compatibility – that is, whether discs recorded on earlier models will be able to be played on future models.
The newer models, called Hi-MD, can upload data directly to (PC) computers. The recordings they produce are high quality, especially for the price (in uncompressed PCM format). There are 1GB discs available, which will record about an hour of audio. The older 74 min discs record about 24 mins 30 sec and the 80 min discs about 27 mins.
Some models have level monitors but they can be hard to read; one of the reasons the devices use little power is that power is conserved in many different ways (by, for example, not backlighting the display screen very much). This can be a problem in very bright or very dark conditions. Furthermore, the recording units need to be kept away well from the microphone. Another reason that they have such good battery life is that the motor doesn’t run all the time. The disadvantage of this is that when it starts you can hear it, and if the unit is too close to the microphone it will be heard on the recording. (There are several ways to solve this – one is to much sure the minidisc recorder is behind the microphone and at least 2’ away from it. Another is to put some obstacle between the unit and the microphone (like yourself; sit slightly to one side of the microphone and have the minidisc recorder on the other side of you from the microphone).
A final issue with minidisc recorders is potential data degradation. Some have reported data corruption in the transfer of the recordings to the hard drive (Klaas 2005) although this can be minimised by restarting the computer before transferring data, and using only the smaller capacity discs. Furthermore, any rewriting on the disc represents a potential for data corruption and degradation of the signal, so it is not advisable to move tracks around on the original disc.
In summary, minidisc recorders have some benefits and some obvious failings. They are a good choice if you cannot afford to buy another type of machine (such as a compact flash recorder) or if battery life and compact size are paramount. The disadvantages can be minimised and the recorders are considerably better than the analogue equivalents for the price.
Compact flash are the most recent type of recorder. They record onto memory cards. They are solid state recorders, which means that they have no moving parts and are thus more durable. Their battery life is acceptable. The recording can be uncompressed raw 16 or 24bit PCM, or compressed mp3. The recordings are easy to transfer to computer with a card reader.
These recorders are quite new and there are different reports about the different models on the market. some models require a particular type of microphone to work well. As yet there is little information about their relative performance, and it seems (so far at least) that all brands have positive and negative reviews. Some think the Edirol R-1 recorder is very good, for example, although others have noted very low input levels for mic inputs (although they are better for line in, so if you have to digitise analogue tapes it is useful). The Marantz PMD670 seems to be a solid and flexible machine, although again the levels seemed to be on the low side for recording and it is not very portable. It is also rather complicated.
 Although having said that, both my field microphones are condenser mics and I have never had a problem with them and humidity. See below, Section 2.5, for information about what sampling rates are.
 Information on this is available from Linda Barwick’s site: http://luddite.cst.usyd.edu.au/pmwiki/pmwiki.php/Linda/InterimDigitalAudioReport
 Ladefoged (2003) has some information about how to set this up.
 Although for some information on this see Nash (2003) (Carlos’ thesis) and XXX (the paper Aoju sent me)