What is Acoustic Beamforming and Why Use it?
Our ears provide us with two different signals of sound. Our brains use this to tell which direction sounds come from and to allow us to focus in on particular sounds coming from particular directions, while tuning out noise. Acoustic beamforming takes a similar approach. By combining the output of several microphones, more directional recordings can be produced: reverberant sounds from the environment and background sources of noise can be reduced. By adjusting the amount of delay from the different microphones in the array, the beam can be steered and pointed in different directions. Our current software detects the loudest signal at any point in time and steers the array towards that signal.
For bioacoustics researchers who need high-quality recordings, acoustic beam-forming can provide a directional output from a passive recorder. Measuring details of bird songs, for example, depends on being able to reliably determine what is reverberation and what is signal. The improved quality of recordings available in Songbeam allows this for a greater proportion of the songs that are recorded. And the animal doesn't need to be singing in front of the array: because the beam is dynamically steered, it will "focus" on whatever the loudest signal is at any point in time.
For users interested in detecting species, Songbeam's beamforming approach also has advantages. By improving recording quality and reducing noise, it can improve the detectability of sounds. "Nocmig" users, for example typically use directional microphones to increase the detectability of faint calls. But such directionality means that only a small area of the sky overhead is covered. Songbeam should allow directional-style detectability, but over a much wider range of the sky.
Songbeam is a growing project and we plan to refine and improve our beamforming. For example, beamforming also allows us to estimate the direction from which sound was produced, adding additional capability and functions to the device. Our current model has been designed for our work on bird song. The spacing of the four microphones in our array means that it works up to a maximum frequency of 10kHz and that the beamforming effect should be less evident beneath 1kHz. If you are interested in analyses over different frequency ranges, please get in touch!
How to Use SongBeam
(1) Assembling the Recorder (barebones version)
The recorder requires:
(a) Teensy 4.1 microprocessor. You can purchase this with or without header pins soldered on. If you choose the "without" option, you will need to solder on header pins yourself. With header pins attached, the Teensy should slot onto the SongBeam board.
(b) Battery holder. SongBeam requires power between 4.5 and 6V. We use 4 d-cell holders, which provides long-lasting power, but 4 AA cells (for example) should work well too. You must connect the battery holder's leads to the + and - terminal of the SongBeam.
(c) Batteries. As well as the main power batteries, you need to connect a CR 2032 coin cell battery to set the time.
(d) Micro-sd card. Faster cards will perform better; we have tested up to 128GB cards. Format the card in exFAT format.
(e) A micro USB data cable (note: many micro USB cables are power only!)
Note: Batteries (excluding the coin cell) should always be disconnected from the board, when the USB cable is used, to prevent the possibility of power flowing back into your computer.
Once assembled, with the coin cell in place, connect the SongBeam's Teensy 4.1 to a computer using the micro USB cable. SongBeam uses software, which can be downloaded from here. You will also need to install the TeensyLoader application. Open TeensyLoader, click File|OpenHex file, then open the downloaded Hex file. Finally, click the white button on the Teensy, followed by the "Program" arrow on the TeensyLoader application. The System should now be ready to use.
(2) Scheduling Recording
To schedule recording, the Teensy needs a file called config.txt loaded on its microsd card. An example of this file is show below with instructions for how to edit it.
(3) Analyzing Recordings
SongBeam will record standard four-channel wav files. Recording time and date is encoded in the file name. The files can be opened and explored using standard acoustic software, such as Audacity. However, to obtain the benefits of beamforming, specific beamforming software will need to be employed to merge the tracks. We have included an implementation of beamforming in our Luscinia bioacoustics software. To employ it, simply add a SongBeam recording as a recording within Luscinia, then in the Controls pane, select "StereoMode" and "1&2&3&4" to merge the four channels.
We also provide a standalone utility to beamform recordings without Luscinia here.
How to schedule recordings with the config.txt files
To get started, use this version of config.txt as a template: copy, paste it into your microsd card and edit it.
DeviceID: the name given to the device. This will appear at the beginning of file names recorded by the device.
RecordStartHrs/RecordStartMins: time of day to start recording in hours and minutes
RecordLengthMins: how many minutes to record for.
FileLengthMins: When recording, SongBeam will record files of this length. Because of the power management chip used in the design, it will restart approximately every 30mins during recording, irrespective of this setting, so not all recordings will be of the same length.
RecMon: - RecSun: Whichever days are shown, the SongBeam will record.