AudioMoth: A low-cost acoustic device for monitoring biodiversity and the environment

Environmental sound is a powerful data source for investigating ecosystem health. To capture it, scientists commonly use ruggedized, but expensive acoustic monitoring equipment. In this paper we fully describe the hardware build of a low-cost, small, full-spectrum alternative, called AudioMoth. The credit-card sized device consists of a printed circuit board, micro-controller and a micro-electro-mechanical systems microphone. This simple to construct device facilitates: (1) deployments in remote locations, with a small size and a simple mechanism that allows it to be retrofitted into numerous low-cost ruggedized enclosures; (2) long-term monitoring, with low-power operation; (3) modular expansion, with easy to access general purpose input and output pins; and (4) acoustic detection, with onboard processing power. 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CCBY license (http://creativecommons.org/licenses/by/4.0/). Specifications table: Hardware name AudioMoth Subject area Conservation biology, environmental, educational tools Hardware type Acoustic sensor Open source license Creative Commons Attribution 4.0 International Cost of hardware 53 USD for 1 unit, 25 USD for 1000 units Source file repository Available with the article 1. Hardware in context The analysis of environmental sound is a rapidly developing branch of ecological research, often referred to as ecoacoustics [1]. The science relies on analyzing stored datasets of captured sound. Datasets are often collated from audio recordings generally taken either on mains powered acoustic setups, or more often by battery powered passive acoustic monitoring (PAM) devices. A full list of current acoustic hardware and software can be found in the WWF guidelines for passive acoustic monitoring in ecology and conservation [2]. Traditionaly PAM is performed using commercial devices, such as the SongMeter 2 A.P. Hill et al. / HardwareX 6 (2019) e00073 series from Wildlife Acoustics (www.wildlifeacoustics.com) and the BAR series from Frontier Labs (www.frontierlabs.com. au). These portable commercial devices are valued for their excellent recording quality, making them well suited to bioacoustics research. Although the cost of these devices is considerably cheaper than complex mains powered setups, they are still an expensive research tool for conservation uses, at a starting price of approximately 800 USD. For many applications requiring coverage of large areas, this cost presents a restriction on usage. In certain projects, such as the use of acoustics to monitor biodiversity or poaching, cheaper units that enable researchers to cover larger areas could have utility even if some sound quality is lost in minimizing unit price [3]. To this end, custom designed, fit-for-purpose solutions have seen a rapid increase over the last 3 years, often facilitated by new low-cost technologies [4]. Accessible and affordable single-board computers, such as the Raspberry Pi, have disrupted the acoustic monitoring market, with many new releases of open-source modular forms of PAM [5–8]. Despite reducing the unit cost to less than 100 USD, modular build-yourself devices are restricted by their high power consumption, sometimes requiring a car battery to sustain longer deployments, and the time required to build each device, often needing hobbyist software and electronics knowledge. Here, we describe AudioMoth: an environmental and wildlife acoustic monitoring tool that addresses many of the barriers associated with pre-existing acoustic monitoring tools. The small size and low power consumption of the device make it amendable to long-term and large-scale scientific research [9]. It provides a manyfold lower cost alternative to pre-existing equipment, and lowers the technical barrier to entry into open-source hardware. This is the first paper to fully describe the AudioMoth hardware build and two types of enclosure to house it. 2. Hardware description AudioMoth consists of a single credit-card sized (58 48 15 mm) printed circuit board (PCB), which includes a sidemounted switch, universal serial bus (USB) port, red & green light emitting diode (LED), and microSD card port. The single board design doubles as an enclosure, with components placed between the board and the battery holder on the top layer of the two-layer PCB (Fig. 1a). The hidden component placement together with an indented switch and SD card connector means the device is robust to knocks that usually occur during general use. Sound is captured through a drill hole, which is located inside the silkscreened microphone symbol on the bottom PCB layer (Fig. 1b). Behind the drill hole sits a bottom ported MEMS microphone. Programming pins (PROG) and four general purpose input/output (GPIO) pins are located on this bottom layer for easy access. The GPIO pins create the option to plug in external modules that interface with the device, allowing users to add hardware modules that extend the device’s functionality. The PROG pins enable the device to enter programming mode, which enables users to upload and update device firmware by USB. AudioMoth’s main advantages over pre-existing tools is its lower cost, lower power usage, small size and ease of use. Its cost, approximately 50 USD for a single unit, is at about 10 times cheaper than pre-existing commercial equivalents. Its size enables easy deployments, with numerous devices able to fit in one field backpack. Its easy-to-use supporting software makes it possible for users at any skill level to configure a device for multiple applications. It can handle both audible and ultrasonic sound capture from one microphone and can be deployed in application specific enclosures. These improvements are achieved through a combination of low-level audio processing that can be optimized for low-power applications, simple single board construction, and a user-centered software/hardware interface. Instead of using a Linux based processor that is harder to optimize for power usage, AudioMoth is built around a power efficient ARM Cortex M4 micro-controller. This micro-controller benefits from floating point functionality, which when combined with low-power operation makes it possible to achieve on-board detection for real-time sound filtering [10]. Fig. 1c shows the hardware overview and the typical flow of operation when running the basic firmware. Users can interface with the device using the three-position hardware switch, and software running on a Windows, Mac or Linux computer via the USB connector. The software provides a graphical user interface (GUI) configuration application, which allows custom parameters within hardware to be changed without needing to modify the firmware. Parameters can be adjusted in the application for multiple deployment scenarios. The switch allows the user to shift between three firmware states, which can be customized using the open-source code. Typically, the default switch position enables the device to start recording immediately with default (if not previously configured) or selected (if previously configured) parameters, the middle switch position enables the device to be configured by USB using the configuration application, and the custom switch position enables the configured parameters to be applied for time scheduled recordings. The analog peripherals and the microphone switch on when the device is ready to record. The analog microphone signal is amplified by the internal firmware-controlled op-amp circuitry. The output of which is then converted to digital samples by the 12-bit analog to digital converter (ADC), which automatically fills a buffer located in the external static random access memory (SRAM) using direct memory access (DMA). Over-sampling can be applied to generate 16-bit samples if required. DMA allows samples to be routed to the external memory in low energy modes without needing to wake up the processor. Using DMA saves substantial battery power during deployments. When the buffer is sufficiently full, it is saved to the microSD card via the serial peripheral interface (SPI) bus. A more detailed description of how the hardware works will be covered in Section 3.1. 1 The device is a revision on a previous model of acoustic logger, AudioMoth1.0.0 (www.openacousticdevices.info). The revision, AudioMoth1.1.0 has improved power management and robustness for outdoor field deployments, using custom built or off-the-shelf enclosures. Fig. 1. AudioMoth. A.P. Hill et al. / HardwareX 6 (2019) e00073 3 AudioMoth can be configured to record at many sample rates, making it suitable for monitoring sounds from different source types. These include: anthropogenic noise, such as gunshots, chainsaws or engine noise (8 kHz sample rate); audible wildlife, such as bird, insect or frog vocalization (48 kHz sample rate); and ultrasonic wildlife, such as bat or amphibian calls (384 kHz sample rate). The device can be used in multiple deployment scenarios, such as scheduled or triggered acoustic monitoring in remote areas, handheld acoustic monitoring, large-scale acoustic monitoring projects, long-term acoustic monitoring projects, environmental monitoring for education, and large scale citizen science projects.


Hardware in context
The analysis of environmental sound is a rapidly developing branch of ecological research, often referred to as ecoacoustics [1]. The science relies on analyzing stored datasets of captured sound. Datasets are often collated from audio recordings generally taken either on mains powered acoustic setups, or more often by battery powered passive acoustic monitoring (PAM) devices. A full list of current acoustic hardware and software can be found in the WWF guidelines for passive acoustic monitoring in ecology and conservation [2]. Traditionaly PAM is performed using commercial devices, such as the SongMeter

Hardware description
AudioMoth consists of a single credit-card sized (58 Â 48 Â 15 mm) printed circuit board (PCB), which includes a sidemounted switch, universal serial bus (USB) port, red & green light emitting diode (LED), and microSD card port. The single board design doubles as an enclosure, with components placed between the board and the battery holder on the top layer of the two-layer PCB (Fig. 1a). The hidden component placement together with an indented switch and SD card connector means the device is robust to knocks that usually occur during general use. Sound is captured through a drill hole, which is located inside the silkscreened microphone symbol on the bottom PCB layer (Fig. 1b). Behind the drill hole sits a bottom ported MEMS microphone. Programming pins (PROG) and four general purpose input/output (GPIO) pins are located on this bottom layer for easy access. The GPIO pins create the option to plug in external modules that interface with the device, allowing users to add hardware modules that extend the device's functionality. The PROG pins enable the device to enter programming mode, which enables users to upload and update device firmware by USB.
AudioMoth's main advantages over pre-existing tools is its lower cost, lower power usage, small size and ease of use. Its cost, approximately 50 USD for a single unit, is at about 10 times cheaper than pre-existing commercial equivalents. Its size enables easy deployments, with numerous devices able to fit in one field backpack. Its easy-to-use supporting software makes it possible for users at any skill level to configure a device for multiple applications. It can handle both audible and ultrasonic sound capture from one microphone and can be deployed in application specific enclosures. These improvements are achieved through a combination of low-level audio processing that can be optimized for low-power applications, simple single board construction, and a user-centered software/hardware interface. Instead of using a Linux based processor that is harder to optimize for power usage, AudioMoth is built around a power efficient ARM Cortex M4 micro-controller. This micro-controller benefits from floating point functionality, which when combined with low-power operation makes it possible to achieve on-board detection for real-time sound filtering [10]. Fig. 1c shows the hardware overview and the typical flow of operation when running the basic firmware. Users can interface with the device using the three-position hardware switch, and software running on a Windows, Mac or Linux computer via the USB connector. The software provides a graphical user interface (GUI) configuration application, which allows custom parameters within hardware to be changed without needing to modify the firmware. Parameters can be adjusted in the application for multiple deployment scenarios. The switch allows the user to shift between three firmware states, which can be customized using the open-source code. Typically, the default switch position enables the device to start recording immediately with default (if not previously configured) or selected (if previously configured) parameters, the middle switch position enables the device to be configured by USB using the configuration application, and the custom switch position enables the configured parameters to be applied for time scheduled recordings. The analog peripherals and the microphone switch on when the device is ready to record. The analog microphone signal is amplified by the internal firmware-controlled op-amp circuitry. The output of which is then converted to digital samples by the 12-bit analog to digital converter (ADC), which automatically fills a buffer located in the external static random access memory (SRAM) using direct memory access (DMA). Over-sampling can be applied to generate 16-bit samples if required. DMA allows samples to be routed to the external memory in low energy modes without needing to wake up the processor. Using DMA saves substantial battery power during deployments. When the buffer is sufficiently full, it is saved to the microSD card via the serial peripheral interface (SPI) bus. A more detailed description of how the hardware works will be covered in Section 3.1.
AudioMoth can be configured to record at many sample rates, making it suitable for monitoring sounds from different source types. These include: anthropogenic noise, such as gunshots, chainsaws or engine noise (8 kHz sample rate); audible wildlife, such as bird, insect or frog vocalization (48 kHz sample rate); and ultrasonic wildlife, such as bat or amphibian calls (384 kHz sample rate). The device can be used in multiple deployment scenarios, such as scheduled or triggered acoustic monitoring in remote areas, handheld acoustic monitoring, large-scale acoustic monitoring projects, long-term acoustic monitoring projects, environmental monitoring for education, and large scale citizen science projects.

Design files
The design files consist of all the needed content to manufacture, assemble and deploy AudioMoth in two variations of weather proof enclosure (Table 1).

Hardware
'Hardware' is a folder containing an Eagle schematic (AudioMoth1-1-0.sch) and Eagle PCB layout (AudioMoth1-1-0.brd). The schematic is split into labelled functional groups: power, audio, memory and debug. The device can be powered from any 3.6 V-20 V DC supply. This power supply connects to a regulator that converts the DC supply to a stable 3.3 V, or alternately the device can be supplied by 5 V USB power, which connects to the internal voltage regulator inside the micro-controller. To reduce the load on the DC supply when USB power is used, a N-channel and P-channel MOSFET array disconnects the supply power when the USB is connected. The power circuitry also has digital noise isolation that uses a series resistor and ferrite bead to isolate the analog audio circuitry from the digital transmission lines (Fig. 2).
The audio circuitry begins at the microphone input, which is routed directly to the micro-controllers analog peripherals (Fig. 3a). Inside the pre-amp block, two on-board op-amps create a cascaded inverting amplifier that can be controlled by the supporting software for gain setting adjustments (Fig. 3b). A low noise DC voltage is generated from an external voltage reference chip to create a stable bias voltage that feeds into the positive inputs of both op-amps.
The micro-controller links to an external SRAM chip, which expands on the internal RAM resources. The external SRAM interfaces with the micro-controller using an external bus interface (EBI) (Fig. 4). Direct memory access (DMA) allows the device to sleep in low energy modes while data is sampled and routed between the analog interface and the external SRAM chip. The SRAM chip is split into an eight element circular buffer in firmware. This buffer structure allows acoustic data to fill in contiguous elements. When a set of elements have filled with a sufficient number of acoustic samples, the microcontroller wakes up to store those elements to microSD card. This is done while simultaneously filling the next element, thus allowing a continuous stream of audio to be recorded to storage. Only secure digital high-capacity (SDHC) microSD cards, formatted to FAT32, are compatible with AudioMoth. SD cards over 32 GB in capacity may require explicit reformatting to FAT32 to function correctly.
The user interface is arranged on the bottom layer of the PCB layout (AudioMoth1-1-0.brd), with all external peripherals marked with clear silkscreen annotations (Fig. 5). These annotations include switch positions, USB port, microSD card port, LED locations, expansion header pin functions and programming header locations. Users interface with the physical hardware via the switch, which is used to define different operation modes. The USB port is used to upload configuration settings from a cross-platform configuration application, where recording schedules and audio requirements can be set from an intuitive GUI. The LED's, which can be viewed from the side of the device, provide feedback to the user as to what state the device is in (www.openacousticdevices.info/led-guide), the microSD card port allows the user to save and transfer audio recordings. The expansion port provides an interface for the potential development of external modules. Fig. 5a displays the microcontroller pin numbers associated with the header pins. These numbers can be used to create expansion firmware for Audio-Moth. The GPIO pins, tx and rx, connect to the micro-controller's universal asynchronous receiver/transmitter (UART) peripheral on pins B9 and B10. Example modules for potential expansion using UART include GPS (www.adafruit.com/product/746) and wireless commutation transmitter/receiver modules (www.pycom.io/product/lopy4/) for internet of things (IoT) applications.

Software
Software is split into two sections: (i) 'Firmware', the software that runs on the device; and ii) 'Applications', the software that supports the device. All source code for these two sections can be accessed on the device GitHub page (www.github.com/OpenAcousticDevices). 'Firmware' is a binary file that can be used to program AudioMoth by USB. A full description of the firmware project, library and example code can be found on the GitHub Wiki page (www.github.com/OpenAcous-ticDevices/AudioMoth-Project/wiki). 'Applications' contains the downloadable Flash application, the Electron based Configuration App and Time App. Three downloadable variations of each application are included for Unix, Mac and Windows desktop operating systems. The Flash application allows AudioMoth to be programmed over USB using a compiled firmware binary through the onboard boot-loader. The Flash application is run from the command line.
The Configuration App enables AudioMoth to be configured by USB. When connected to AudioMoth, the Configuration App displays the device's onboard time, identification number (ID), firmware version and battery voltage. The Configuration App allows users to set the audio parameters and recording schedules for various deployment scenarios (Fig. 6). Four separate recording periods can be set during a 24hr period, with the option between eight different sample rates (8 kHz, 16 kHz, 32 kHz, 48 kHz, 96 kHz, 192 kHz, 256 kHz, 384 kHz), five different gain levels (low = 27.2 dB, medium-low = 28.7 dB, medium = 30.6 dB, medium-high = 31.6 dB, high = 32.0 dB), any sleep duration less than 12 h, any recording duration less than 12 h, and LED's can be enabled or disabled. After configuration, the application calculates and displays the daily memory and battery consumption. The Time App enables AudioMoth to be configured by USB. When connected to AudioMoth, the Time App displays the device's onboard time, identification number (ID), firmware version and battery voltage. The Time App only allows the time to be set, and is useful to use with customized firmware. Table 2 presents the software source locations.

Enclosures
We will illustrate two simple laser cut versions of AudioMoth enclosure. The first case design is located in the 'AA enclosure' zip file, which contains three CAD design files. These files are used to laser cut acrylic and neoprene gasket sandwich layers. They can be assembled to form a waterproof case for AudioMoth when using a through hole AA battery connector. The top and bottom acrylic layers (Fig. 7a) use 5 mm acrylic and the two middle layers (Fig. 7b) use 10 mm acrylic. Neoprene gaskets (Fig. 7c) are used between the layers to prevent water ingress. Many off-the-shelf enclosures can also be used to protect AudioMoth too. Examples of these can be found at www.openacousticdevices.info/support/enclosures.
The second case design is located in the '6 V enclosure' file, which contains three CAD design files and a hardware schematic and layout file. The hardware files form a base PCB assembly to hold AudioMoth onto a connector for the 6 V battery. The CAD design files are used for laser cutting acrylic sandwich layers to form an internal structure to hold the PCB assembly against a 6 V battery. Most of the internal structure uses 3 mm acrylic (Fig. 8a) held together by cable ties. The internal structure is designed to fit within common industrial 110 mm diameter coupling drain pipes. One 5 mm piece of acrylic is used to hold AudioMoth flush against the internal drain pipe plug end (Fig. 8b). Fig. 8c shows a jig paper cutout that can be used as a guide for the microphone drill hole and the placement of the rectangular key in Fig. 8a. The small semi circle structure in this figure is the rain hood, which should be used on the outside of the drain pipe plug to prevent moisture pooling.

Bill of materials
The bill of materials are split into the AudioMoth hardware and two version of custom enclosure, the AA battery operated version and the 6 V battery operated version. Table 3 includes unit pricing to purchase components sufficient to build a single device or a batch of 1000. Batch ordering components and PCBs in this way results in a larger cost saving per device.

Build instructions
This section describes the build instruction for the PCB and two waterproof enclosures for outdoor deployments. The AA battery enclosure is to be used for deployments requiring less battery capacity and the 6 V battery enclosure is to be used for deployments requiring large battery capacity.

AudioMoth PCB
The bare PCB or the assembled PCB can be acquired by uploading the bill of materials (Table 3) and the 'Hardware' design files, or generated Gerbers to any PCB assembler. PCB assemblers usually charge a tooling setup cost, ranging from 200 USD to 1000 USD to place components. Ordering small batches of assembled devices from PCB assemblers is not recommended due to this high setup cost. Manually assembly from the bare board would be more cost effective. However, the components would need to be soldered by hand. Manually soldering components takes approximately 4 h per board. In terms of both cost and time, the most effective route to acquire an AudioMoth is through group purchase websites, such as www.groupgets.com/manufacturers/open-acoustic-devices/products/audiomoth. This acquisition route delivers assembled and programmed AudioMoths for a fixed unit cost of 49.99 USD. If not assembled, the components can be placed using the following instructions: 1. Sort the AudioMoth bill-of-materials into labelled component bins (Table 3). 2. Solder components to their silkscreen designator locations, which are displayed on the PCB bottom layer (Fig. 5b). Most components can be soldered using a soldering iron. For components smaller than '0402' sized passives such as Y2 and U2, a heat gun, or oven should be used. Solder paste and a stencil can help with manual assembly using an oven.

List of equipment
For soldering PCB (if not pre-assembled) Soldering iron or oven/heat gun Solder/solder paste Tweezers Flux pen Flux off Alcohol PCB cleaner

AA battery enclosure
The AA battery enclosure uses the default AA battery holder to power the device. Once the battery holder is attached, the unit can be placed inside an enclosure. This acrylic enclosure can be laser cut using the design files or ordered from any laser cutting service. The enclosure can be built using the following instructions:   Table 4) to the PCB. 4. Once soldered and solder is cool, trim the battery connector legs flush with the PCB using wire cutters. 5. Lay out all of the components to make the acrylic enclosure (Fig. 9c). 6. Place the first 5 mm acrylic cutout flat on a surface, with the four hex spaces inserted into each hex slot. 7. Place the first expanded neoprene gasket through the four hex spaces so it sits flush against the first acrylic layer. 8. Place the first 10 mm acrylic cutout through the four hex spaces, with the SD and USB protrusion located in the top left corner. 9. Place the second expanded neoprene gasket through the four hex spaces so it sits flush against the second acrylic layer, with the SD and USB protrusion located in the top left corner. 10. Place the second 10 mm acrylic cutout through the four hex spaces, with the SD and USB protrusion located in the top left corner.  11. Place the third expanded neoprene gasket through the four hex spaces so it sits flush against the third acrylic layer with the SD and USB protrusion located in the top left corner. 12. Place the configured AudioMoth inside the acrylic case so the SD card lines up with the protrusion in the 10 mm acrylic cutouts. 13. Place the final 5 mm acrylic sheet on top of the enclosure and make sure the drill hole lines up with the microphone. 14. Screw up the four top thumb screws (Fig. 9l). 15. Turn over and screw the last four thumb screws into the bottom of the enclosure. 16. Firmly screw up all thumb screws and stick the acoustic vent over the microphone hole. 17. Thread the cable ties through the drill holes on the back acrylic layer.

List of equipment
For enclosure build Laser cutter Hot glue gun  Hot glue sticks Wire cutters Expanded neoprene gaskets from recommended distributor 5 mm and 10 mm acrylic sheets from distributor, pre-cut or in sheets

6 V battery enclosure
The 6 V battery enclosure uses two PCBs in its construction: the main AudioMoth PCB and a separate additional PCB that acts as a power connector to support the larger 6 V battery to power the device. The additional PCB can be acquired by uploading the '6 V enclosure' design files, or generated Gerbers to any PCB assembler. This enclosure uses cable ties to hold the acrylic layers in position. The acrylic layers can be laser cut using the design files or ordered from any laser cutting service. This case is useful as it extends the operation life of AudioMoth, with some 6 V batteries having a capacity of 24Ah (http://data.energizer.com/pdfs/529.pdf). The 6 V battery must have two-spring connectors and not the screw post terminals. (Table 5).
The enclosure is ruggedized and waterproof for hostile deployments. The 110 mm drain enclosure can be purchased at local hardware stores worldwide. This availability means the enclosure does not have to be shipped to the deployment country. The enclosure can be built using the following instructions: 1. Solder the two small single pin connectors to the AudioMoth battery connector pins and cut flush with the bottom layer of the PCB (Fig. 10a). 2. Place and lineup the 5 mm acyclic cutout over the 6 V enclosure PCB (Fig. 10b). 3. Place two small dabs of hot glue on the top layer PCB as shown previously in Fig. 9a, making sure not to glue on top of any components 4. While the glue is still hot, place the AudioMoth PCB onto the 6 V enclosure PCB and flip the assembly over so the AudioMoth PCB sits flush with the workshop surface (Fig. 10c). 5. Once the glue has cooled, solder the AudioMoths power pins to the 6 V enclosure PCB keeping the AudioMoth PCB bottom layer flush with the top of the 5 mm acrylic cutout (Fig. 10d). 6. Once solder is cool, trim the battery connector legs flush with the PCB using wire cutters. 7. Lay out all of the internal acrylic structural components (Fig. 11a). 8. Thread the cable ties through the first 3 mm acrylic cutout (Fig. 11b). 9. Line up the square notch cutouts, place the second 3 mm acrylic layer over the battery, and thread the cable ties through (Fig. 11c). 10. Line up the square notch cutouts, place the third 3 mm acrylic layer with PCB assembly over the battery, and thread the cable ties through (Fig. 11d).  11. Insert the battery while cable ties are loose and push the complete PCB assembly firmly over the battery and tighten the cable ties (Fig. 11d). Make sure to line up the battery connector pads and battery springs correctly. 12. To prepare the external enclosure to hold the internal structure place the printed drill jig into one end of the plugs (Fig. 12b). 13. Using the hot glue gun, glue the rectangle key into position (Fig. 12b). This is used to lock the internal structure into position. 14. Drill a 5 mm diameter microphone hole using the same printed jig as a guide (Fig. 12b). 15. Place double sided tape or hot glue onto the underside of the 3 mm acrylic rain hood and place over drill hole (Fig. 12c). 16. Insert and lock the internal structure into position making sure the microphone lines up with the drill hole. Place the acoustic vent on the outside of the enclosure over the drill hole.

Operation instructions
Fully illustrated operation instructions are available with this article. Maintained operation instructions can be found on the AudioMoth website (www.openacousticdevices.info). For up to date instructions visit (www.openacousticdevices.info/ getting-started).

User instructions for firmware development
AudioMoth firmware is written in C and can be edited and compiled using the free Silicon Labs IDE, Simplicity Studio (www.silabs.com/products/development-tools/software/simplicity-studio). For detailed instructions on how to setup Simplicity Studio and create an AudioMoth project, visit the AudioMoth wiki at www.github.com/OpenAcousticDevices/  AudioMoth-Project/wiki/AudioMoth. Firmware can be uploaded to the device by building the project and using the compiled '.bin' file via USB.

Validation and characterizations
In this section, we specify how to validate the correct build of the AudioMoth hardware and measure its performance. Since the hardware is built around a micro-controller, validation simply involves checking the functionality of the peripherals used. The peripherals include, the USB boot-loader, the USART SPI interface that communicates with the microSD card, the EBI, the op-amps and the ADC.

Boot-loader validation
To test the boot-loader, the Flash application (Section 3.2) can be used. Validation of the boot-loader is achieved through the following steps: 1. AudioMoth's micro-controller has a boot-loader pre-programmed at factory. When put into programming mode, a fully assembled AudioMoth will be recognized by a PC when plugged in via USB. 2. To put the device in programming mode the PROG pins must be shorted as the device is powered on. A metal paper clip can be used to do this. 3. While the PROG pins are shorted, the USB connector can be plugged into the device. The short on initial power up puts AudioMoth in boot-loader mode. 4. The Flash application (Section 3.2) can then be run to validate boot-loader functionality. For this test, start the 'Flashing by USB' instructions ( Table 6) at step 6. 5. The device port number should return on the command line Flash application. Follow the rest of the 'Flashing by USB' instructions from step 6 to program the hardware with the 'AudioMoth-Firmware-Basic' firmware. If boot-loader has failed the flash application will return NULL, meaning either the PROG pins were not shorted correctly at start-up, or there is another hardware fault.

Peripheral validation
Validation of all other peripherals can be achieved by using a programmed AudioMoth to make a WAV file voice recording to microSD card. Voice recordings can help quickly identify hardware issues. For example, skips heard in the recorded voice indicate a slow SD card, or incorrect voice pitch indicate problems with sample rate. For validation, it is useful to download the free open-source, cross-platform audio software, Audacity (www.audacityteam.org/download) to view AudioMoth recordings. Validation of the peripherals is achieved through the following steps: 1. To record, insert batteries and set the switch to DEFAULT. The red LEDs should immediately start flashing to represent a recording is taking place. If both green and red LED flash together an SD card error has occurred. 2. Quietly count down from ten about 1 m from the AudioMoth device during the red LED flashes. 3. After the count down, the switch should be set back to USB to stop the recording. 4. The microSD card can now be taken out and inserted into the PC. 5. If the USART SPI peripheral is functioning correctly a WAV file should have been created, named '19700101_xxxx.WAV'. 6. To validate the functionality of the EBI, op-amps and ADC peripherals the playback of the counting sequence needs to be analyzed. The recording can be opened in Audacity. 7. Voice should be clearly heard without distortion during playback. Recordings can be viewed as a waveform showing amplitude against time (Fig. 13a), and also as a spectrogram, showing the spectrum of frequencies against time (Fig. 13c).

Measuring power
Keeping power consumption low is critical for long-term acoustic monitoring applications in remote areas. To measure the current consumption on AudioMoth, a Silicon Labs starter kit can be used (www.silabs.com/products/developmenttools/mcu/32-bit/efm32-giant-gecko-starter-kit). Using the energy profiler application within the Simplicity Studio IDE, the starter kits VMCU pins on the external header can be used to power AudioMoth. For detailed instructions on how to use the Starter Kit for power measurements visit the following forum post: www.silabs.com/community/mcu/32-bit/knowledge-base.entry.html/2014/05/21/using_aem_to_measure-BdWl. AudioMoth consumes approximately 30uA while sleeping and draws 10 mA to 40 mA during recording, depending on sample rate. Table 7 shows how current changes with sample rate using the SanDisk Extreme 32 GB microSD card. These are guideline numbers as each model of micro SD card may draw different overall current during a recording.

Typical applications and deployments
Acoustic monitoring has multiple applications, including the monitoring of specific species [11][12][13][14], soundscape analysis [15][16][17], environmental surveillance [18,19], and as a hobby for wildlife enthusiasts [20]. Each application can change the requirements for hardware configuration. The most important sound characteristic to capture is the range of sound frequencies emitted by the target. To record the source frequency the sample rate must be at least twice the top end frequency pro- Description of LED flash meanings www.openacousticdevices.info/led-guide duced by the target. Half the sample rate is called the Nyquist frequency, any sound above the Nyquist frequency will not be recorded. Sample rates must therefore be set to correspond to at least double the sound source in question. Hardware configuration guidelines for different applications and the corresponding source frequencies are shown in Table 8. The speed of SD card required depends on the sample rate used, with higher sample rates requiring higher SD card speeds. The speeds required for each application can be found in the last column of Table 8. As well as source frequency, other device parameters can be adjusted for certain applications. The duty cycle routine and timed recording schedules can be matched for specific target activities. For example, bat echolocation calls are active at sunset. The duration of sound can be used to set a duty cycle routine, which saves the amount of power and memory required for each deployment. Sounds that are known to be frequently emitted, or to last longer can have larger sleep periods between each recording event. Short duration sounds, or sounds less frequently emitted will need smaller sleep periods. For night deployments or deployments prone to theft, indication LEDs can be turned off to prevent unwanted attention. Gain can be changed to adjust for different background noise levels during recordings. Generally, gain can be left at the default medium setting unless a high noise environment is known.  If this is the case gain should be lowered accordingly to prevent audio clipping or distorting. Sample rates, time schedules, duty cycle routines, LED function and gain can all be set using the Configuration App. Battery life often outperforms memory capacity when using a 32 GB SD card; the penultimate column shows how often the memory card should be swapped during the life cycle of one set of 3Ah batteries.

Deployment instructions
The deployment of battery powered devices is an important, and often problematic, part of acoustic monitoring in remote locations. Before deploying AudioMoth, first it must be configured with the appropriate recording schedules and acoustic parameters to perform its task. The battery and memory consumption predictions should be recorded to estimate battery life, this can be used to plan a collection schedule to replace batteries and memory cards. Before placing devices in the pre-built enclosures the switch must be put into CUSTOM mode.

Conclusion
In this paper we have described the hardware design of the AudioMoth acoustic logger. We have given instructions on the location of its open-source design files, the construction of two enclosures to house the PCB, validation processes required after hardware assembly and guidelines on how to deploy the final working product. AudioMoth can be used for many different acoustic monitoring applications, such as monitoring ultrasonic bat calls, and audible wildlife vocalizations. Audio-Moth's low cost, small size, low power operation and simple construction increases the scalability of deployments in remote areas. This enables bigger conservation research questions to be answered than has been possible up to now with conventional acoustic technology.

Declaration of Competing Interest
None.