PDA

View Full Version : DC Offset Removal Questions and Plug-Ins


RenderBot
03-29-2007, 10:33 AM
I'm currently dealing with a bunch of files that require DC offset and thought I'd try and get some answers from you all. Before getting into optional solutions I'd like to ask the following:

1. From my understanding, DC Offset is a mean amplitude which resides above 0dB. I've always assumed that this offset is introduced by the preamp and or phantom power. Is this always the case?

2. If not, what other sources of power and or circuitry can be responsible.

3. I also considered the possibility that a condenser microphone, once charged, could be responisible. This leads me to the question, is a condenser microphone a passive or active device when charged? I'm assuming that though it requires power, it's still an active device as it puts its own stamp on the signal being introduced. What about other mic types?

4. Assuming that elements (Microphone, preamp, phantom) remain the unchanged, is the deviated from zero absolute or variable. By the looks of my recordings it seems to vary, however, different files, of course, have varying levels of low frequency information.

5. Can preamps, mics, phantom power sources, etc. be recalibrated to solve this problem? Would this be something that would fall under warranty?


Plug-Ins:

1. I've tried to use the free Pro Tools DC Offset plug-in and have never had any luck with it. I'm also unaware of how its attempting to solve the problem. Does anyone know if it's attempting to notch out a certain frequenc(y)(ies)? It doesn't seem to be applying a high-pass function, however this is how I've always gone about resolving the problem.

2. As stated above, I've always used a High-pass filter to fix DC Offset. Does anyone know if there a plug-in that actually adjusts the position of the waveform itself? DC offset is something that affects the mean level of the entire waveform, therefore, couldn;t a plug analyze the offset and reposition the waveform to sit at a new user definable point? Maybe this already exists?

Cheers,
Frank

Richard Fairbanks
03-30-2007, 04:48 AM
1. From my understanding, DC Offset is a mean amplitude which resides above 0dB. I've always assumed that this offset is introduced by the preamp and or phantom power. Is this always the case? A DC offset that matters is when there is a steady voltage value that is recorded even when there is no audio signal. That is the waveform that is a straight line above or below the "zero" mark. It can cause a click when you edit it. Usually caused by the AD converter. Nearly all commercially available analog inputs, whether mic or line level, include capacitors that block straight DC voltage (and low frequencies below 5 or 10hz also), so even if your favorite mic pre has some DC offset it is blocked when you plug it into your 192io, for instance. Technically, a blocking capacitor is actually introducing a variable DC offset while blocking a steady offset, but that is a relatively small tradeoff and just clouds the discussion.

2. If not, what other sources of power and or circuitry can be responsible. More likely introduced by poorly designed or broken AD converter, or the circuits just ahead of the AD converter.

3. I also considered the possibility that a condenser microphone, once charged, could be responisible. This leads me to the question, is a condenser microphone a passive or active device when charged? I'm assuming that though it requires power, it's still an active device as it puts its own stamp on the signal being introduced. What about other mic types? Not a concern because of answer 1. By the way, ALL active (condenser) mics must block DC offsets, either with a capacitor or a transformer. It is how they separate the supply power from the signal.

4. Assuming that elements (Microphone, preamp, phantom) remain the unchanged, is the deviated from zero absolute or variable. By the looks of my recordings it seems to vary, however, different files, of course, have varying levels of low frequency information. Variable. It is relatively common to see voice that contains more energy in the positive wavefronts than negative, or vice versa. Less common with traditional instruments, but certainly exists. While the mean value of these examples may not be zero, I don't consider this actual DC offset, but automatic detection methods of DC offset can be fooled by it

5. Can preamps, mics, phantom power sources, etc. be recalibrated to solve this problem? Would this be something that would fall under warranty? Many of them cannot. Rarely you will find an adjustment (older equipment sometimes, new rarely), usually it is circuit design that eliminates DC offset. Capacitors block DC but also affect linearity and phase response of low frequencies. Servo circuits can be thrown off track by low frequency information, so that they add low frequency distortion or they don't work right. Lots of loud kick drum hits, for instance, can fool them.

1. I've tried to use the free Pro Tools DC Offset plug-in and have never had any luck with it. I'm also unaware of how its attempting to solve the problem. Does anyone know if it's attempting to notch out a certain frequenc(y)(ies)? It doesn't seem to be applying a high-pass function, however this is how I've always gone about resolving the problem. I am fairly certain it calculates a mean value of the region you have selected, then adds that value to the waveform. It does not hi-pass, but it can be fooled by some waveforms. It has often worked very well for me.

2. As stated above, I've always used a High-pass filter to fix DC Offset. Does anyone know if there a plug-in that actually adjusts the position of the waveform itself? DC offset is something that affects the mean level of the entire waveform, therefore, couldn;t a plug analyze the offset and reposition the waveform to sit at a new user definable point? Maybe this already exists? I prefer to use hi-pass for realtime mixing. It is quick, easy to do, and non-destructive. Look for the lowest frequency you can find. 10hz is better than 20. 2hz is good if you could find it in a plugin.

cmaynes
03-30-2007, 10:25 PM
The Dc filter in ProTools and Sound Designer was a high pass filter. It did not measure and offset. I think it runs at 5 Hz.


charles maynes

Frank Kruse
03-31-2007, 05:35 AM
I'm currently dealing with a bunch of files that require DC offset and thought I'd try and get some answers from you all. Before getting into optional solutions I'd like to ask the following:

1. From my understanding, DC Offset is a mean amplitude which resides above 0dB. I've always assumed that this offset is introduced by the preamp and or phantom power. Is this always the case?





Hi Frank. Nothing can be above 0dB in a digital system. DC-offset can also be caused by bad A/D-converters. A DC-offset is a signal with frequency=0 at a certain amplitude (the offset you see in the waveform).

As stated before simply apply a high-pass with a very low frequency. And the offset is gone. It has the same result like offsetting the waveform.
A plugin that actually shifts the waveform would have to subtract the off-set from every sample. For this purpose it would have to know the offset at any given time (every sample). This would only work easily if the offset is constant over time. But the result will be the same. A very low low-cut is much easier.

Also see here

http://www.harmony-central.com/articles/tips/eq_dc_offset/

Frank.

Richard Fairbanks
03-31-2007, 06:16 AM
Hi Charles, are you taling about the audiosuite plugin? Check it again and I think you will discover that my original post is correct!

Try these two tests:
Create a very low frequency sinetone and process it for DC offset. I used Digi's audiosuite Signal Generator to make a 1 second 20hz sinetone and then pitch changed it using an audiosuite pitch shifter, allowing the length to change. Drop the pitch -24 semitones (2 octaves down, the limit of digi's shifter), the file is now 4 second long and is 5hz. Drop it again -24 semitones, the file is now 16 seconds long and is 1.25hz. Process it using digi's DC offset audiosuite plug. There is no significant change in DC offset or level of the tone. This proves that hi-pass is not used by digi's plugin. If it was a hi-pass filter the tone 's level would be lower.

For the second test, create a short file of silence. You can use the Signal Generator plugin again, just set the output level to infinity. Now zoom in close enough to use the pencil tool and draw a waveform entirely above (or at least mostly above) the centerline. Make the drawing as long or as short as you wish. When you're bored with drawing, just select the area you drew plus a little more that is still zero signal. Process with the DC offset plugin and notice what happens, the entire waveform will be shifted down including the area that was formerly zero. Undo it and process the same area with a hi-pass filter and you will notice a radical difference in results.

One method is not better than the other depending on circumstances, and the results nearly always sound identical, but technically the results are different. They both fail by some degree at times and they can both cause signal clipping that was not present in the original. I am glad to have a choice. I would like to see Digi include both options in the DC offset tool.

cmaynes
03-31-2007, 12:43 PM
I will give your test a try, but I will say that when I was at Digi (I was there for about 4 years), I asked about this and the answer I received was as I had mentioned it is a hpf- which was a simple filter as far as DSP designer was concerned.- now it is possible that the filter is not steep enough, but for me that is specutation.

I did just test your method, and found that even down to less than 1 Hz- if bipolar modulation is present the signal will not be identified as DC offset- If you were to grab a portion of the relatively steady signal while it is away from the zero crossing point you will find the DC offset Plugin will function as expected. I did this on a section of the a single cycle which was 60 feet in duration.

I do agree that DC tools such as the old Apogee plugin would be nice- The current ones are sort of cheap and dirty- It seems this is an opportunity for someone out there...

charles maynes


Hi Charles, are you taling about the audiosuite plugin? Check it again and I think you will discover that my original post is correct!

Try these two tests:
Create a very low frequency sinetone and process it for DC offset. I used Digi's audiosuite Signal Generator to make a 1 second 20hz sinetone and then pitch changed it using an audiosuite pitch shifter, allowing the length to change. Drop the pitch -24 semitones (2 octaves down, the limit of digi's shifter), the file is now 4 second long and is 5hz. Drop it again -24 semitones, the file is now 16 seconds long and is 1.25hz. Process it using digi's DC offset audiosuite plug. There is no significant change in DC offset or level of the tone. This proves that hi-pass is not used by digi's plugin. If it was a hi-pass filter the tone 's level would be lower.

For the second test, create a short file of silence. You can use the Signal Generator plugin again, just set the output level to infinity. Now zoom in close enough to use the pencil tool and draw a waveform entirely above (or at least mostly above) the centerline. Make the drawing as long or as short as you wish. When you're bored with drawing, just select the area you drew plus a little more that is still zero signal. Process with the DC offset plugin and notice what happens, the entire waveform will be shifted down including the area that was formerly zero. Undo it and process the same area with a hi-pass filter and you will notice a radical difference in results.

One method is not better than the other depending on circumstances, and the results nearly always sound identical, but technically the results are different. They both fail by some degree at times and they can both cause signal clipping that was not present in the original. I am glad to have a choice. I would like to see Digi include both options in the DC offset tool.