RememBird - spectrograms with the internal mic, with MKE300 and compared to HiMD PCM

The aim of this comparison is to see how RememBird and RememBird-X with MKE300 differ in the spectrogram, and to see what sort of differences there would be compared to a HiMD recorder. Although the comparison is mainly about the sonic results, in terms of ease of field operation the RememBird wins hands down. With the Sony MZ-NH700 HiMD deck you have to press buttons about six time, and look at the LCD screen before you are set in manual record mode, because the recorder defaults to AGC recording. With RememBird you press the on button for two seconds, then you are ready to record with a single button press. You also have a preroll buffer with RememBird.

The subject - a thrush on a tree above a quarry, about 20m away (estimated).. Wind was low, esimated < 5mph, and did not trouble either mic. RememBird tells me the recording was made at 7:37:30 on the 15th Feb 2007. RememBird recording bitrate was set to 320kbps.

Figure 1: Raven image of internal mic RememBird (L) and RememBird-X using MKE300 (R).

Song Thrush recorded using RememBird (distance about 20m) gain scaled by about 22dB (original unscaled)

Song Thrush recorded using RememBird-X and MKE300 (distance about 20m) gain scaled by about 28dB (original unscaled)

Song Thrush recorded using HiMD and MKE300 (distance about 20m) - same bird, same mic and bird location but not the same snippet as I only had one MKE300. This recording was made in PCM and compressed to 320kbps MP3, the same bitrate as the others.

HiMD

A comparison was also made of the RememBird-X recording at 320kbps with a HiMD recorder recording in PCM. This was performed by replaying a previously recorded swallow, via the test fixture to adjust to mic levels. This was chosen to be able to replicate the input signal accurately. It is therefore, not strictly representative of what happens in real life. If you mouse over the pictures you can compare with the original (top two) and RememBird-X with the HiMD (bottom). Raven was set to the default settings (50,50,256) for all of these, and signal levels were normalise on the WAVs going into Raven. The lateral shift is because I did not manage to exactly match the time scale and zoom on Raven. 

Signal levels from the test fixture (as used in the technical measurements) were set to -20dBFS for the RememBird (approximately -48dBu, 300µVrms), which is broadly where it operated on most of my subjects. The same level was used for the HiMD.

Fig.2 HiMD versus original signal on mouseover

Both rerecordings are more noisy, and the extreme HF detail above 19kHz is not visible in the noise. Around 6.75s one of the harmonics is lost on the RememBird between 12 and 14kHz, and the one below is weakened. This corresponds to a high signal level and it does seem as if this higher harmonic is beig discriminated against by the MP3 coding. However, to put this into perspective, all of these are clearly the same song-phrase and nothing seems to have be added to the signal by the MP3 coding. There has often been controversy about the use of lossy compression in bioacoustics, however, this is the best example of a difference that I could find.

This is an mp3 version of the original swallow sound, recorded with a Telinga parabola at 2m at Ennerdale

more RememBird recordings

© WSRS 2007

The inclusion of any item on any page of this website, is based on an individual's comments and experiences, and are not an endorsement by the Wildlife Sound Recording Society.
The opinions expressed are those of the individual authors and not necessarily those of the Wildlife Sound Recording Society.