A Quick-and-Dirty Comparison of (Some) Handheld Digital Audio Recorders

I'm a sucker for handheld digital recorders, having accumulated a small army of them since good quality models started hitting the market a few years back. Conventional wisdom (backed by laboratory measurement) is that the microphones and preamps on these devices range in quality from worse than you'd expect to surprisingly good, sometimes even approaching the fidelity of studio-grade gear.

I'm interested in outdoor field recording of ambient soundscapes, so aside from portability and usability, the qualities I'm most concerned with in these devices are low noise, microphone fidelity, and the character of the stereo image. Some of these properties can be measured numerically, but there's also a subjective element best captured by comparing actual recordings made with different devices under the same conditions.

So I made a bunch of outdoor recordings of the same subject at the same time with different recorders. The subject is simple and dull: the ambient sounds heard in suburban backyard on a quiet weekday afternoon, including dogs, children, birds, trains, cars, airplanes, and so on. This is about as entertaining to listen to as you might expect, but the environment approximates the ambient conditions under which many outdoor soundscape recordings are made. I may make more recordings of different subjects and in different places in the future, but for now, the peculiar music of American suburbia will have to do. This is a fairly undemanding test. Recordings made in quieter environments, for example, would make the self-noise of the various preamps and capsules more evident.

The seven recordings are all edited to 2'12" long, and are in level-normalized MP3 format. They are simultaneous but not precisely synchronized (you can't sensibly listen to the left channel of one and the right of another). No EQ or other post-productions effects were applied, except for level normalization. I tried to set the gains to approximately the same level across the recorders. and the loudest sounds (the dog bark) peaked at between about -6dB and -16dB in the raw recordings. All captures were made at the highest quality settings of the devices, which in most cases was 96KHz at 24 bits PCM. For recorders with a choice of microphone configuration, I used coincident XY cardioid at 90 degrees. There's also a mystery microphone thrown in to the mix, plus a mid-side pair of Sennheiser MKH-800s (wide-cardioid and figure-8) recorded on a Nagra VI. Yes, that's not exactly small and handheld, but it's a useful baseline for comparison.

The recordings were made in a mild breeze. I covered the included foam windscreens with Rycote "Windjammer" furballs, which was mostly effective in eliminating wind noise as a variable here.

It's worth noting that the cost of the equipment being compared here spans a roughly 17dB price range (assuming, of course, that money is power). So this isn't exactly a fair fight. Indeed, perhaps the most remarkable thing to discover from this test is that even relatively inexpensive recorders can sound pretty good, all things considered.

The clips here are all in 48KHz MP3 format, converted via Logic Pro 8. The files were compressed to 320 Kbps and are about 5MB each.

• Nagra ARES-MII. "Green band" clip-on XY microphone. Recorded at the device's maximum quality setting of 48KHz/16bit. Rycote Windjammer.

  Audio: Suburban-ARESMII.mp3.

• Olympus LS-10. Fixed pair of 90degree cardioids, spaced at 4.5cm. Covered with Rycote Windjammer for Nagra ARES-M

  Audio: Suburban-LS10.mp3.

• Sennheiser MKH-800 Mid-Side. Mid set to "wide cardioid". Mics mounted inside a Rycote blimp + windjammer. Recorded on Nagra VI.

  Audio: Suburban-MKH800.mp3.

• Sony PCM-D1. Equipped with aftermarket furry windscreen.

  Audio: Suburban-D1.mp3.

• Sony PCM-D50. Mics set to 90 degree near-conicident position. Equipped with aftermarket furry windscreen.

  Audio: Suburban-D50.mp3.

• Zoom H4n. Mics set to 90 degrees. The Rycote windjammer for the NT4 fits the windscreen well.

  Audio: Suburban-H4n.mp3.

• Mystery Microphone. XY configuration. How does it compare?

  Audio: Suburban-XY.mp3.

Update 20 July 09: Rob Danielson created a very nice QuickTime visualization of snippets of these samples here.

Update 21 July 09: I did a second set of comparison recordings with a more challenging stereo subject: small waves breaking against the beach on a moderately windy afternoon. This is a tough subject because waves and water produce what is essentially white noise across the sound stage, and so an effective stereo image requires not only clear channel separation, but also the capture of subtle sonic detail amid the laterally spread noise.

The microphones were set up about 4m (12 feet) from the edge of the San Francisco Bay, on a deserted beach near Richmond. (The wind protection of the mystery mic proved inadequate, and so that recording is omitted here).

Here, the stereo image is much flatter than in the suburban example, effectively rendered as monaural on some of the recorders.

As before, these 48 second clips have been level-normalized and compressed into (320Kbps) MP3s, but are otherwise as they came out of the box, with no EQ or other post-processing applied.

• Nagra ARES-MII. "Green band" clip-on XY microphone. Recorded at the device's maximum quality setting of 48KHz/16bit.

  Audio: Sea-ARESMII.mp3.

• Olympus LS-10. Fixed pair of 90degree cardioids, spaced at 4.5cm.

  Audio: Sea-LS10.mp3.

• Sennheiser MKH-800 Mid-Side. Mid set to "wide cardioid". Mics mounted inside a Rycote blimp + windjammer. Recorded on a Nagra VI recorder.

  Audio: Sea-MKH800.mp3.

• Sony PCM-D1.

  Audio: Sea-D1.mp3.

• Sony PCM-D50. Mics set to 90 degree near-coincident position.

  Audio: Sea-D50.mp3.

• Zoom H4n. Mics set to 90 degrees.

  Audio: Sea-H4n.mp3.

Bottom photo: Microphones at the beach.


Entire contents copyright © by Matt Blaze. Some rights reserved. The audio clips are available under a Creative Commons Attribition/Noncommercial/Share-Alike license .

mab blog here...

Home page here...