I didn’t get what you’re recording. What-U-Hear points to something else. If you can’t record the work, for example, on an internet site, where did you get the undamaged clip from? The story doesn’t hang together. Koz

When I turn up the gain on the Sample.mp3 I get the impression that I am hearing the music mixed with something like a faint radio broadcast (i.e. spoken words) but I cannot make out any of the words. If you have an original MP3 why are you re-recording it? Why not simply File…Import it into Audacity.


You’re getting “drop-outs” (sometimes called an “xrun”)
Basically it is a discontinuity in the waveform that occurs if some part of the system is unable to keep up with the flow of data. Each time it happens you will hear a click. Here is a zoomed-in close-up of one of the clicks:
Unfortunately there is no single answer to the problem. The job is to try and locate the data bottleneck and speed it up.
Common causes include: other programs or background processes hogging the CPU, the hard drive being unavailable due to other programs/processes reading/writing when there is audio data to be written, fragmented hard drive (Windows only), shortage of available RAM or drive space, DMA not enabled for the hard drive (not common on modern computers), USB hubs (if using a USB sound card), inefficient or buggy sound-card drivers…
See here for suggestions of how to approach resolving the problem: http://wiki.audacityteam.org/wiki/Managing_Computer_Resources_and_Drivers


I use much lower spec. hardware than that, so that’s not the problem.

My best guess would be the sound card drivers. “Sound Blaster X-Fi” have a terrible reputation for drivers - not for gaming, apparently they’re great for that, but frequently buggy for recording. Check on the Creative website and check that you have the latest drivers for your operating system.


I used to use Soundforge on a lesser machine than this, and had ZERO issues recording.

We all have zero troubles recording. In the old TV Repair Business, this would be a “Tough Dog.” Only one failure like it and multiple people have tried to fix it. Alternately, there’s nothing broken, but it doesn’t work.

I found the discontinuities in the waves, too. There are small slivers of time that are simply not represented in the waveform. They’re gone. That gives you the razor sharp vertical lines – and the tick.

If drivers don’t do it:

Can you make it worse? What can you do to make this kind of recording collapse into a pile of garbage? Load on the applications and memory hogs and see if it gets worse.

Change something major. Start peeling off parts of the machine until the problem changes. Peel off the drives. Unmount everything. Have you ever run the Microsoft Memory Tester? It’s free and it will point to multiple different difficulties, not just memory.



Oh, and you should see if you can make it fail without “What-U-Hear.” What U Hear is an artificial construct that runs play and record at the same time. It’s like trying to troubleshoot a ball of yarn. Yes, by all means, plug your iPod into the AUX or Line-In, record it and see if it ticks then. That’s a much more straight-line, simple path. Koz



One of the items on our hit list is the ability to change the problem.

– Make it better.
– Make it worse.
– Make it change.

All are valuable.

I need to peel off and go to work, but I can tell you the problem sounds exactly like running a USB microphone through a USB hub along with the keyboard and mouse. One of the Systems people called into a video conference at the company and he sounded exactly like that. Random ticks throughout his voice as the system struggled for breath. Koz

Same thing. That chop is the missing fragment in the Stx song. Koz
Screen shot 2012-03-27 at 8.44.18 PM.png

Describe the architecture. How are the drives connected to the machine? Koz


Back in the dim memory of analog systems, over a 50% range of damage, you got a 50% range of perceived distortion in the work. The digital people assured us that digital systems either worked or they didn’t. Turns out digital system have their problems, too. They can dance rapidly between total success and total failure, and they’re much more difficult to service when they do that because a small fraction of percent change in conditions causes an enormous change in damage.

Today, your system seems to be working properly. Did the humidity go up? Power coming out of the wall socket slightly higher this morning? This is the kind of thing that drives knife-edge problems to change effect.

That Microsoft Memory Tester…


… is instructive because of the way it’s used. It’s not a one-pass test. You set it to go through its suite of different memory examinations (dripping faucet, running puppy, checkmate, etc.)and then you tell it to loop and repeat the entire suite all night. Then you go to bed. If the machine makes it through eight or twelve hours of constant pounding under slightly different conditions of heat, humidity, line voltage fluctuations, etc. etc, then it’s probably good to go. I’ve discovered memory stick failures, power supply failures, and I killed off a motherboard once – all hours into the test.

So that’s why it’s instructive not only to make it better, but to be able to make it worse, to make sure you didn’t just gently nudge the problem on the good side of the knife edge and it’s going to come back next Thursday – when you really need the recording.

Just out of interest, how many megabytes-per-second am I looking at during a 48000Hz/24 Bit recording?

It won’t make any difference to you because Windows doesn’t allow 24-bit recording (AFAIK), but as a reference (and I haven’t had enough coffee to figure it out) recordings like this easily travel down a USB2 connection with a little left over. What’s that, 360Mb/Sec, half-duplex? Some remarkably slow and sloppy computers run Audacity just fine, so you have a bottleneck in there somewhere…


24 bit = 3 bytes
Stereo = 2 channels of samples
48kHz 24 bit stereo = 3 x 2 x 48 kB per second.

That does not sound like much, but the important point is that it must be continuous sustained throughput.
USB can go very much faster than this, but USB 1 and 2 are not ideal protocols for continuous throughput, they are much better at short fast bursts of data.

and that was after rebooting the system?