Choosing A Recorder for Frog Calls

First off, let me start with the caveat that I am not a sound engineer or audio junkie. I am a herper who has taught himself a little bit about recording amphibian calls and wanted to share some of what I have learned through trial....and lots of error.

There is lots of great information about recording wildlife online, including advice on which recorders are useful for these tasks.  Wildlife recorders face some challenges that studio recordings and concert recordings including very quiet environments with quiet subjects.  This isn't the typical problem that people encounter recording a rock concert, for example.   Because of this, the sensitivity required by wildlife recording brings out the flaws in some otherwise useful recorders.


Recording calling frogs and toads adds another level of challenge that normal wildlife recorders don’t face. It is usually done in wet areas, sometimes hip-deep in water, usually in the dark, and often in the rain. You might be miles from your home or vehicle and have to carry everything with you.  And then you might have to stand/crouch for 10-15 minutes in the dark waiting for the darn frog to start calling again!


So how do you choose a recorder for this task?


Of course, if budget is no concern and sound quality is the foremost consideration, you will probably end up with a high end field recorder such as the those made by Sound Devices  and expensive microphones with the appropriate wind and weather protection . But for most herpers with just a casual interest in documenting frog calls, that is probably overkill and it won’t fit in your pocket.


So how do you choose?


Here are some variables to consider:


-------------------------------------------------------------------------------------------------------------


Frequency Range


First, some technical mumbo jumbo.  Hertz is the scale used to measure the frequency or pitch of a sound.  High pitched sounds are higher in frequency and have higher Hertz values.  For some scale, the lowest note on a piano is 27.5 Hz and the highest note is 4186 Hz.  The low C (C2) note sung bass singers is around 65Hz while the high C sung by a good operatic soprano is around 1046Hz.


These values are sometimes expressed in Kilohertz (KHz).  1KHz = 1000Hz.  


The generally quoted boundaries for human hearing are ~20Hz at the low end ~20,000Hz (20Khz) at the high end.

Sample Rate

Hertz can be used to express the frequency of the sound being generated, but confusingly, it is also used to measure the sample rate (or recording rate) of a recording.  These are not the same. 


The sample rate is a measure of how often the recorder samples the sound per second.  You can think of it as analogous to how a video captures motion.  A video doesn't actually capture motion, it captures a whole series of still shots (frames) of the action and by playing them back at a particular speed they appear to be continuous motion.  

The recording sample rate is similar.  The recorder is not measuring the sound continually, but is measuring the sound over and over again and these individual captured sounds are then played one after the other to give the whole continuous sound.  Sample rates are very high with most modern recorders sampling the sound at rates of 48Khz (or 48,000 samples per second).  Some more specialized recorders support sample rates of 96Khz, 192Khz and even 384Khz.  These recorders captures sounds well above the level of human hearing but can be useful in documenting some insect or bat calls.

Why do I need to know this stuff?   Because there is a theoretical limit we need to worry about.  A recording can only capture sounds up to a frequency that is approximately 1/2 of the sample rate.

So using a sample rate of 8Khz will only capture sounds up to a frequency of ~4Khz.  If you use a higher sample rate (say 48Khz), you will be capturing sounds up to ~24Khz frequency.

CD Quality recording is 44.1Khz which means CD quality recordings can capture sounds up to 22,050 Hz (22Khz).  Most humans have a hearing range that tapers off near 20Khz at least when they are young.  Older people have a hearing range that tapers off at 12Khz or even lower.  (My hearing is failing, so I don't hear much above 11.5Khz).  But because "CD quality sound" includes is sampled at 44.1Khz, it can include all the ranges capable of being heard by humans.

Lastly, larger sample rates means larger file sizes.  

OK, enough background already!......

So the obvious first thought might be to turn to a Voice Recorder or use an app on your cell phone or tablet since they are inexpensive and readily available. This can work in some circumstances but voice recorders or the voice recorder apps in many phones are optimized for capturing the human voice.  As we just saw, the typical frequency range of the human voice is usually between 80 and 2000 Hz..  Therefore a recording rate of 4KHz is usually enough for voice and and 8KHz sample rate will capture almost any sounds a human can make.


The frequency cutoff also depends on the particular mode you use in your recorder. For example, one common entry level voice recorder made by Olympus only records frequencies up to 3 kHz in certain space saving modes.   Why limit the frequency range in voice recorders?  Simple.  Higher recording rates = larger file sizes.  Large file sizes means you can't fit as many hours of recording on the recording medium and if you read the ads for voice recorders their main selling point is usually how many hours they can store.


So voice recorders are great for recording human voices, but once we step up record natural sounds many animals call outside of this range. Several species of frogs in the genus Eleutherodactylus, for example, have very high pitched calls that can be in excess of 8 kHz. The North American Little Grass Frog (Pseudacris ocularis) calls are in the 7.5 kHz range.  To capture those, you would need a recorder that records at at least a 16KHz recording rate.


To hear the difference, here's a recording of a chorus of frogs from Puerto Rico including the lower pitched calls of the Common Coqui (Eleutherodactylus coqui) and the Red-eyed Coqui (E. antillensis).  Over the top of this lower pitched background, you can hear the high-pitched whistle of the Whistling Coqui (E. cochranae) as it calls three times. The Whistling Coqui call has a peak frequency of 4100Hz.  (This recording is at 16KHz so it includes everything below 8Khz)





Now here is the same recording saved as an 8KHz recording as it might be picked up by an  inexpensive voice recorder.





See, no more Whistling Coqui!  It's call is just above 4Khz so it was not picked up in the voice recorder (emulated) recording.



Here are those two recordings in a row with the 16KHz recording followed by the 8KHz recording and the spectrogram for those recordings.  You can see (highlighted) the high pitched call of the Whistling Coqui and the fact that it is "missing" from the second 8KHz recording.








Even frogs whose carrier frequency (main note frequency) is below the 8 kHz threshold, the actual “sound” of their call is dependent on higher sideband or harmonic frequencies which can be well above the carrier frequency. 


A good example of a North American anuran call in which the sidebands/harmonics influence the overall sound of the call is the Ornate Chorus Frog (Pseudaris ornata). The following is a recording of a Pseudacris ornata taken from somewhere out on the world wide web (sorry, I don’t remember where!). I took a short section of the recording and copied it. The first time it plays, it is playing at “CD quality” sample rate of 44.1 kHz (with a maximum frequency of 22 kHz). It then repeats having the sample rate reduced to 8 kHz (max frequency of 4 hKhz). Here is a sonogram of what is in this recording. The highlighted calls in the first recording are all that is left in the second. All the higher parts of the call are lost.



You can hear the difference the loss of these over tones makes:





When you listen to the recording, the second time through, it sounds distinctly different. This is due to the loss of the harmonics/sidebands above 4 kHz. So if this frog had been recorded with a low end voice recorder at a high compression setting, the call would sound like the second part of the recording. While that is certainly enough to identify the species in question, it clearly loses some of the texture or tone of this particular species’ call.


So a dependable frog recorder needs to be able to capture the range of frequencies used by most anurans so your recordings be representative of what the frog sounded like in the field. I would generally want a recorder capable of 48Khz recording rate which would record sounds up to 24Khz.  There are anurans known to call above these frequencies, but those ultrasonic frequencies are above the range of human hearing anyway so probably aren’t of interest to most casual anuran recordists. If you were interested in ultrasonic calls, there are recorders that go well beyond this. Some recorders can capture ultrasound with sampling rates of 96kHz, 192kHz or even 384kHz but they generally require specialized ultrasonic microphones to do so. While these recorded “sounds” can’t be heard by the human ear, they can have their frequencies brought down into human hearing range after recording. This is how bat biologists record and analyze bat calls. 


-------------------------------------------------------------------------------------------------------------


Signal to Noise Ratios


A microphone works by picking up the vibrations in the air generated by the sound wave and generates a weak electrical current corresponding to that sound wave. However, that current is so weak that in order for the signal to produce an audible recording, it has to be amplified before it is recorded. This is called “preamplification” and most recorders have built in “preamps” for this purpose.


This introduces a new problem. The microphone converts the sound to a weak electrical signal and the preamp the increases the strength of this electrical signal.  But during the process of amplifying this signal, it is possible to add extraneous electrical "noise" to the amplified signal.   This electrical noise shows up as hiss or other distracting background noise in the final recording.  


Recordists describe the quality of a good microphone and preamp by their signal to noise ratio.  In other words, how much of the desired subject (signal) is present in the final recording when compared to the extraneous noise added by the amplification process.


Well designed preamps will minimize this noise and amplify mostly the "signal" that we are interested in but poorly designed preamps can add excess noise.  Less expensive recorders often have lower signal to noise ratios than more expensive recorders, but there are good entry level recorders with good S/N ratios.


When dealing with loud sounds like guitar music, drums, people talking close to the microphone, etc., it isn’t an issue as the signal is so loud that the additional preamp noise is inconsequential. But when the quieter sounds of the natural world even a small amount of added noise can really interfere in the ability to discriminate quieter frog calls.   So the wildlife sound recordist wants a microphone and recorder combination that has a high signal to noise ratio.


Here is a comparison of the two species of anurans (Hyla chrysoscelis and Incilius nebulifer) recorded simultaneously from the same spot with two different recorders. One is recorded with an older model Motorola Android cell phone and the other with an Olympus LS-11 Digital PCM recorder. The Olympus recorder has good preamps and produces less noise than the Android phone. You can actually see this in the sonogram for this recording.





On the top recording (the phone) the background is much darker. This dark background represents noise in the recording. In the bottom sonogram, the background is much ligher while the frog calls are still as dark. Therefore you can see there is more signal (the darkness of the frog calls) compared to the noise (darkness of the background).


Here is a shorter section of these recordings played one after the other. In the first part, you hear the phone recording followed by a second of silence then the recording made by the Olympus LS-11. Listen to the background hiss in both recordings and compare how well the calls stand out. It is easiest to hear in the short buzzy trills of the Cope’s Gray Treefrogs (Hyla chrysoscelis). The longer trill is a Gulf Coast Toad (Incilius nebulifer).





It is worth pointing out here that there has been a significant improvement in the quality of recording that you can achieve with cell phones as I outlined in a more recent recording here -
http://frogcalls.blogspot.com/2015/03/recording-with-phones-2015.html 

-------------------------------------------------------------------------------------------------------------

Recording Format


Another consideration is the format in which the recording will be captured. In voice recorders, a selling point is often the maximum number of hours of recording which can be stored in the recorder. In order to maximize that number the recordings may be compressed into lossy formats and the frequency compressed into the range expected for the human voice. This means higher frequency calls may be lost or significantly degraded in the final recording. When choosing a recorder, it is preferable to have a recorder which will save the file in an uncompressed format (aiff, wav) rather than a compressed format (mp3).


There is a nice comparison showing the limitations of voice recorders here - http://www.wildlife-sound.org/equipment/newcomersguide/usenospeechrec.html


-------------------------------------------------------------------------------------------------------------


Other Practicalities


Beyond the technical specifications we need a recorder that is small, field hardy, and somewhat weather-resistant. Furthermore, it is preferable to have a recorder that is easy to adjust and monitor in the field. Some recorders require you to go down through menus to make simple changes like the input level (gain) of the microphone. That can hard to do when you are out in the field, knee deep in water in the dark. Also, since you might be holding a microphone in the other hand, it can be helpful to be able to make changes with one hand.


Storage media and connectivity


How the recorder stores its recordings is another important consideration. While older handheld recorders relied on cassetes or microcassetes, the noise generated by those recorders made them obsolete with the advent of digital storage. Some recorders store their recordings on internal flash memory while others rely on removeable media such as SD cards. Some older models use compact flash cards and others rely on internal hard drives. These methods are OK, but not as field hardy as more modern methods.


Getting the recordings off the recorder is easy with most modern recorders. They either have removeable cards which can be put into a computer and/or they have mini USB ports plugs on the side to allow direct connection. Either way, getting recordings into your computer is a breeze.


Microphones


Another variable to consider when purchasing a recorder is what types of external microphones it will accept. There are two primary microphone plug types used in recorders. Inexpensive microphones and recorders use a 3.5mm TRS plug similar to a headphone plug on an MP3 player. These plugs as small and easy to use. There are inexpensive adapters and extension cords available for these sized cords at almost any electronics stores and even many Wal-mart type stores. Microphones with these types of plugs either require a battery in the microphone for power or rely on the recorder to provide a low voltage plug-in power than can power small microphones directly.


Professional microphones generally have a 3 pin XLR type plug instead. These plugs are larger and most XLR plug microphones depend on power to be supplied by the recorder itself. This recorder-based power supply is often called phantom power. Some recorders are capable of providing phantom power and some aren’t. Phantom power voltages vary from microphone to microphone but most recorders that supply phantom power can supply it at various voltages.

So why would you go the trouble of using an phantom powered XLR type microphone when a 3.5mm microphone would be easier? 


One difference is the sturdiness of the connection. TRS pins can become unplugged easily if pulled. I have on more than one occasion been recording with my 3.5mm plug microphone only to find out it wasn’t plugged in to the recorder and what I was actually recording with was the internal microphones of the recorder. TRS 3.5mm pins are also fairly thin and I have bent a couple when bumping (or dropping :-() my recorder when out in the field. XLR type connections are sturdier and often lock into the socket. In order to be pulled out, you have to depress a pin as you pull so accidentally pulling one out is more difficult.


The other difference is the quality of the sound. When a microphone has its own internal power supply and unbalanced inputs like many TRS 3.5 mm cables, that increases the chance that electrical interference will be produced in the line and show up on the recording as noise. With good microphones and cables this can be reduced, but it is never as quiet as a balanced phantom-powered, XLR type connection. The problem with noise in a recording is that you don’t notice it until you hear a recording that has less.


Part of the learning process is learning to hear the difference between a good (quiet) recording of an amphibian and a bad (noisy) one. Try going to online resources like Inaturalist.org's audio observations and listening to some of the frog and other recordings on there. You will hear a profound difference in the quality based on the different recorders, microphones and techniques used. 

You can also hear this difference by listening to the differences in the recordings used as vouchers in online databases like HERP, Herpmapper.org, and Xeno-canto.org (for birds).

But most importantly, get outside, record some amphibians and have fun! 

And don't forget to upload your mp3 vouchers into a citizen science database somewhere like those mentioned in the last paragraph.  That way your recordings can live forever!

© Chris Harrison


No comments:

Post a Comment