Hi Everyone,
Just some updates.
Changes (since Dec 2012):
The Tape2PC was returned due to terrible SNR, lack of basic functions, and overall poor quality. It was replaced by a regular cassette deck (TEAC W-890R). Thus, there was no USB input so I had to use something else.
Around the same time Audacity 1.3.12 under Wine started exhibiting same problem as higher versions, crashing the system out sound (when Audigy is used). So, I had to look into other ways to making Audacity work.
Things tried
Since Audacity didn’t like Audigy I reconsidered the necessity for Audigy’s front I/O port, bought a couple of adapters and decided to use the PC case front audio ports. So, out comes Audigy and the on-board audio gets enabled (mobo ASUS P5Q3). Installed latest Audacity from Packman repositories (2.0.2) and no more system sound crash, all seemed fine, until tried recording. The “Line in” input presents a significant DC offset; what’s worse is that there is a difference between channels too. And yes, I know that 1st thing that comes to mind is “Normalize” which indeed removes the DC offsets; the problem is the sound activated recording becomes useless. Audacity sees the DC offset as noise, so in my case the noise signal level reported was approx. -36dB on L channel and -42dB on R channel; setting sound activated recording level to -35dB drops some portions of the signal on R, and setting to anything under -36dB records all the time since L has always 'signal". I need “sound activated recording” so I could leave the cassette player do its thing for 3 hrs (auto reverse + continuous play x 2 x 90min) and don’t have be there to stop the recording. I couldn’t find any SW solution, Audacity “suggestions” lists compensating for DC offset prior to recording but was rejected. I booted to Windows 7 and saw the same offset in Audacity for Windows. I concluded the sound controller may have some problem and decided trying another sound card (3rd!).
Got an ASUS Xonar DS; this was OK as price, 7.1 channels and compatible with the Antec PC case audio header. I installed the card, disabled the on-board, and found out this too has DC offsets on the input line, only different values than before. Booted in Windows and was surprised to find there was no offset. Back to Linux and found out the sound driver (oxygen) has a lot of options and settings; one of them is the High Pass Filter. Once this was enabled there is no DC offset on Line in - all good but … when Audacity starts it disables HPF! In other words, you can only enable HPF after Audacity was started; I figured that shouldn’t be a problem creating a small script to launch Audacity and then enable HPF. Moving on, I tried using Audacity and ran into other issues; the device list had about 150 devices. It seems that all those options and settings I mentioned above, are confusing Audacity into thinking they are separate mixer channels; for example there was “attack time” with discrete settings ranging between 1ms and 48ms - each setting was seen by Audacity as a channel. Also, the sound card was detected 3 times, each with a slightly different name. At this point, I decided to return the Xonar card and try something else.
Since Audacity problems with Audigy seem to happen only when Audigy is default system sound out, I reinstalled it and kept the on-board sound enabled. From KDE sound settings I selected default sound out the motherboard sound (speakers connected to it) and default sound in the Audigy (this configuration also allows using Audigy’s I/O port for sound in). This seems to be a working configuration for all applications, (even Skype uses the devices correctly); there is no DC offset on Line in (sound activated level is at -48dB). 
Another working configuration, without Audigy - while tinkering with the above, I went ahead and ordered a Xitel INport Hotwire ($15). Linux recognizes it “out-of-the-box” as USB sound in, no DC offset, no detectable added noise. I was able to record with Audacity using 48KHz sampling rate.
Conclusion(s)
I believe I found at least a couple of issues with Audacity for Linux:
- Audacity doesn’t always “understand” the sound configuration correctly; latest example, the Xonar with 150 devices and I think similar problems happen with Audigy too.
- Audacity does alter system sound settings; in my case see the HPF setting on Xonar (or other reports with Skype problems after Audacity). I think this is what happens when it crashes the Audigy drivers, by forcing incorrect settings, after incorrect reading of the system configuration.
A couple of suggestions for DC offset issue - it could help if there are separate audio record level settings for each channel, or some sort of “zero-out” button to use as reference the current level.
Also, sorry to hear there are no developers for Audacity Linux, wish I was better at this because is sad to see a great Linux application becoming Windows-only.
Thank you and Best Regards!