Snowflake USB Microphone on Linux – with sample files

Today I picked up something called a Snowflake USB microphone by Blue in hopes of improving the sound quality of my screencasts. And while I haven’t had a chance to fully test it, I can say two things: First, it’s a very significant improvement on the sound I was getting from my Logitech USB headset. Second, it is not going to be without its challenges, at least for now. Let me explain.

Let’s hit the problems first. As I’ve seen on several posts on, I seem to be having this weird behaviour where the microphone captures sound fine, but when the audio is played back, it’s at half-speed. That’s only a click or two away from fixing in Audacity. I post-process my screencast audio in Audacity anyway so that’s not a deal-breaker for me. I’m not sure what the problem is, but I know I’m not the only one experiencing it. Unfortunately I have yet to find a fix for it (granted I haven’t really done much of a search yet). I’m running Crunchbang 8.10 and love it, so moving to a newer kernel isn’t an option, at least not immediately.

The other problem I have is that RecordMyDesktop (run either from the gui or commandline) does not play well with this mic. I get a repeated ‘broken pipe’ message when recording and the audio is choppy and almost indecipherable (and at half-speed as noted above). Definitely unusable. I am thinking of recording the audio separately (but at the same time) using the Gnome Sound Recorder. It seems to record the audio perfectly well (despite the half-speed problem). I’m hoping I can just launch the audio recording alongside a video-only recording by RecordMyDesktop. I always end up splitting the audio from the video anyway during my post-processing. The only additional challenge here is making sure everything is in sync in the final product.

The sound quality improvement over my Logitech USB microphone is remarkable. While there is still some background noise, when I remove it in Audacity I don’t get that underwater, over-processed effect. The sound is also much warmer and fuller than what the headset was giving me. Not a big surprise I guess. I think it will be a significant improvement to my screencasts. The microphone comes mounted to a bracket which sits on top of my laptop screen (with protective rubber pads to prevent any marking of the laptop). I was wondering about how the distance vs sound relationship would work, but I have to say that at normal speaking volume, having the mic at that distance seems to sound pretty good. The levels are a little low, so I may have to rig up an impromptu mic stand using my mini Manfrotto tripod or something if I’m not happy with the placement. The build quality is good and it comes with a nice little case that can hold the usb cable along with the microphone itself.

Now I’m just hoping that recording the audio separately comes off as easy as it sounds. Time will tell, but I will definitely try a dry-run in the next few days to see how much harder the workflow is.

The only other major issue I have with my screencasting (besides coming up with good ideas and finding time to do them) is this dang mini-sized MS mouse that I use. It works fine, but the clicks seem terribly loud. Damn you Microsoft! Maybe a Logitech would be quieter. I’m either going to have to hunt down a new mouse with much more silent clicking or take this little baby apart and see if I can quiet down the switches somehow – if I don’t break it first. ;)

I’ve linked to two sample audio files I recorded. Both files were recorded using the Snowflake microphone mounted on my laptop screen about 16 inches away from my face directly in front of me. Both files were amplified slightly to raise the levels using Audacity (and sped up to eliminate the half-speed problem). The first file is the raw file, the second file is the one where I’ve removed some of the background noise.

3 thoughts on “Snowflake USB Microphone on Linux – with sample files

  1. When recording audio with Audacity, I had a problem similar to your problem; audio playback would be slow. The easiest workaround to the problem was to use mhwaveedit to record the audio. YMMV

  2. It appears to be a bug in the driver. According to the Ubuntu bug tracking site, the problem is solved in the latest upstream kernel (see posting #21). Unfortunately, since you’re running Crunchbang 8.10 you’ll probably have to compile a new kernel from source to solve the problem.

Leave a Reply

Your email address will not be published. Required fields are marked *