How much DC offset is reasonable?
Jul 3, 2008 at 8:42 PM Thread Starter Post #1 of 13

trains are bad

Headphoneus Supremus
Joined
Aug 31, 2005
Posts
2,218
Likes
11
I've been wanting to record some vinyl, but I when I do I get a lot of DC offset in the files. I know some is typical but how could I be getting this much?

Originally the DC offset didn't seem so bad, but I was getting clipping, so I added a resistive network attenuator I had lying around, which dropped the levels dramatically, and also seems like it made the DC offset worse. This is a picture after zooming in.

Do you think this could be from my reciever tape outs, or is it my ADC in my soundcard? If I was to put a film capacitor in the line to see if it's from my receiver, what value should I use? If that doesn't fix it, do I just have to get another soundcard?

Sorry I mad the pic so little. It's zoomed up, but the dc offset seems to be about .07, with the scale set to linear. Probably it seems worse because I have attenuated the levels so dramatically; I will build a better, adjustable attenuator later.
 
Jul 3, 2008 at 9:31 PM Post #2 of 13

dfkt

Headphoneus Supremus
Joined
Jul 20, 2005
Posts
4,352
Likes
115
What sound card are you using?

No, "some offset" is not typical. A DC offset of zero is what you should aim for. Any decent professional sound card should provide that. I got two Echo cards and one M-Audio, and offset has never been an issue with those.

Some software like Sound Forge or Wavelab can counteract DC offset... but if you have to use that you already know your sound card is not top notch.
wink.gif


Just get a good sound card instead of tinkering with capacitors and such.
smily_headphones1.gif
 
Jul 4, 2008 at 2:58 AM Post #3 of 13

b0dhi

Headphoneus Supremus
Joined
Oct 12, 2005
Posts
2,070
Likes
21
Since the DC doesn't look like it's been attenuated with the input signal, I suspect it's being added by something down the chain from the resistor network.
 
Jul 4, 2008 at 5:44 AM Post #4 of 13

infinitesymphony

Headphoneus Supremus
Joined
May 15, 2006
Posts
4,621
Likes
15
That's a large amount of DC offset, and speakers could potentially be damaged with it. Try running Audacity's DC offset removal feature. Use Ctrl+A to select the entire waveform, then go to Effect -> Normalize. Make sure that the DC offset is the only one checked and click OK.
 
Jul 4, 2008 at 5:55 AM Post #5 of 13

Redo

Headphoneus Supremus
Joined
Aug 29, 2006
Posts
2,294
Likes
11
Quote:

Originally Posted by infinitesymphony /img/forum/go_quote.gif
That's a large amount of DC offset, and speakers could potentially be damaged with it. Try running Audacity's DC offset removal feature. Use Ctrl+A to select the entire waveform, then go to Effect -> Normalize. Make sure that the DC offset is the only one checked and click OK.



Yup, and if you do want to normalize the audio, use Amplify. Normalize will do left and right independently of eachother in Audacity.
 
Jul 4, 2008 at 6:34 AM Post #6 of 13

Apocalypsee

500+ Head-Fier
Joined
Apr 20, 2007
Posts
517
Likes
12
Talking about DC offset I tested something today. I compared electrolytic vs film (polyester) on DC offset value. With electrolytics I measured 5mV on the output while with film of the same capacitance gives mere 0.2mV. Film sounded much better, cleaner and more detail
 
Jul 4, 2008 at 6:47 AM Post #7 of 13

infinitesymphony

Headphoneus Supremus
Joined
May 15, 2006
Posts
4,621
Likes
15
Quote:

Originally Posted by Redo /img/forum/go_quote.gif
Yup, and if you do want to normalize the audio, use Amplify. Normalize will do left and right independently of eachother in Audacity.


Yeah, Audacity's UI choices for normalize/amplify are awful. It's the only audio program I can think of where the 'Normalize' feature has a hard limit of -3 dB, and the 'Amplify' feature offers the functionality that's disabled in Normalize. Combine that with the lack of adequate bit-rate conversion and other programs (like Wavosaur) start to look much more attractive.

I never knew about the independent channel normalization. That's virtually inexcusable in an audio program.
 
Jul 4, 2008 at 7:56 AM Post #8 of 13

Redo

Headphoneus Supremus
Joined
Aug 29, 2006
Posts
2,294
Likes
11
Quote:

Originally Posted by infinitesymphony /img/forum/go_quote.gif
Yeah, Audacity's UI choices for normalize/amplify are awful. It's the only audio program I can think of where the 'Normalize' feature has a hard limit of -3 dB, and the 'Amplify' feature offers the functionality that's disabled in Normalize. Combine that with the lack of adequate bit-rate conversion and other programs (like Wavosaur) start to look much more attractive.

I never knew about the independent channel normalization. That's virtually inexcusable in an audio program.




I think they changed to this in the newer versions. I found out after I normalized my Allman Brothers - At Fillmore East Vinyl, which entire sides have the drums panned almost entirely on one channel. After normalization, I noticed the right channel stayed fairly similar, yet the left channel was boosted HUGE. Until I found out that Amplify is what I wanted, I was using dbpoweramp's normalizer. It's made me go back and re-process all my Vinyl rips.


What's the deal with the bit-rate conversion?
 
Jul 4, 2008 at 9:41 AM Post #9 of 13

infinitesymphony

Headphoneus Supremus
Joined
May 15, 2006
Posts
4,621
Likes
15
Quote:

Originally Posted by Redo /img/forum/go_quote.gif
What's the deal with the bit-rate conversion?


I did some digging around I guess it's passable, but the main problem is that most of the options are completely hidden away. IMO, bit-rate and dither type should be selectable on the Export screen. As it stands, they're hidden away in the Preferences and 24-bit is not a default export format, it's only selectable under 'Other...' So, I guess it works, but the idea that you'd go into a program's Preferences menu in order to change an important file-type option every time is a little unfriendly, especially since the sampling rate is in a clearly-visible location. Coordinating a conversion from 24-bit/48 kHz to 16-bit/44.1 kHz takes much longer than it should, and it's not easy to spot if it was done correctly.

Then again, I'm used to Bias Peak.
tongue.gif
 
Jul 4, 2008 at 3:16 PM Post #10 of 13

trains are bad

Headphoneus Supremus
Joined
Aug 31, 2005
Posts
2,218
Likes
11
Quote:

'Normalize' feature has a hard limit of -3 dB


No it doesn't. It's set like that by default, but I can easily change it with a simple slider. Maybe you had an old version; I'm using whatever is in the Hardy reps.

I don't have a proper soundcard at all, I'm using the line in on my Biostar 7050 HTPC. The only thing else I have around is an old soundblaster. I don't really want one either since I use digital out for everything else. I suppose it would by silly, but I could use my RH-1 to record to PCM, then use that to output to the computer, since I would rather do that than attempt to actually offload the files anyway.
 
Jul 4, 2008 at 7:24 PM Post #11 of 13

infinitesymphony

Headphoneus Supremus
Joined
May 15, 2006
Posts
4,621
Likes
15
Quote:

Originally Posted by trains are bad /img/forum/go_quote.gif
No it doesn't. It's set like that by default, but I can easily change it with a simple slider. Maybe you had an old version; I'm using whatever is in the Hardy reps.


The Linux version may be different, though I haven't checked for a new version in a while. Here's what I see in 1.2.5:

AudacityNormalize.png


Edit: Well whaddya know, the new Windows beta version (1.3.5 beta) has fixed this problem and 0.0 dB is now the default Normalize value. It's also easier to change the bit-depth (Options under Export). So most of my complaints have already been addressed.
biggrin.gif


But Redo is right. I just tested it, and even in the 1.3.5 beta, Normalize is channel-independent, meaning it will ruin the stereo image on most tracks.

Before normalization:
norm_before.gif


After normalization:
norm_after.gif
 
Jul 4, 2008 at 11:03 PM Post #12 of 13

trains are bad

Headphoneus Supremus
Joined
Aug 31, 2005
Posts
2,218
Likes
11
I'll have to remember that bit about normalize. I didn't know it was channel independent. But I wouldn't go so far as to say it will ruin the stereo image. I can't think of any recordings that intentionally have more average volume on one channel than the other. And if it did, such as if my cartridge was off, a lot of people would expect that to actually be taken care of during the course of 'normalization'.

What's the cheapest linux-compatible soundcard with a useable line-in? Is that AV710 for $20 still around?
 
Jul 5, 2008 at 4:33 AM Post #13 of 13

Redo

Headphoneus Supremus
Joined
Aug 29, 2006
Posts
2,294
Likes
11
Quote:

Originally Posted by trains are bad /img/forum/go_quote.gif
I'll have to remember that bit about normalize. I didn't know it was channel independent. But I wouldn't go so far as to say it will ruin the stereo image. I can't think of any recordings that intentionally have more average volume on one channel than the other. And if it did, such as if my cartridge was off, a lot of people would expect that to actually be taken care of during the course of 'normalization'.


Just use Amplify, it won't do each channel independently. And like I said, normalizing the channels independently with At Fillmore East by the Allman Brothers had a enormously negative effect.
 

Users who are viewing this thread

Top