Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Is this true about the foobar EQ (or digital EQs in general)? (Read 5220 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Is this true about the foobar EQ (or digital EQs in general)?

Hi all,

I recently had someone tell me that digital volume controls were detrimental to the sound because they reduced the "bit depth" when the volume was turned down through them. Is this true?

Conversely, I understand that the EQ feature in fb2k is a digital EQ - will using it have any detrimental effect on my sound along the vein of the aforementioned "loss of bit depth"?

I'm currently streaming FLACs via ASIO to an outboard DAC, then to a headphone amp and then my headphones, and have done some pretty major EQ in just one area (bass) to complement my current setup, which is absolutely beautiful sounding except for the bass department.

Frequency response graphs for this particular headphone show a low bass rolloff of ~8dB at 55Hz, so I'm EQ-ing the opposite to "negate" this. The whole setup sounds lovely, but I'm wondering about the abovementioned "loss of bit depth".

Help!

Is this true about the foobar EQ (or digital EQs in general)?

Reply #1 –
No one?

Surely there must be someone here who is able to answer this?

Is this true about the foobar EQ (or digital EQs in general)?

Reply #2 –
From what i know is that, if you increase volume (using EQ in this case) and it doesn't clip. you don't lose anything. but i maybe wrong.

But.. if i were you, If "The whole setup sounds lovely" then i wouldn't bother about "loss of bit depth".

Is this true about the foobar EQ (or digital EQs in general)?

Reply #3 –
There are many different algorithms for creating volume controls and filtering digitally. Some are good, some are not so good. Although I can't answer the question for you, it is possible. However, because foobar is designed with SQ in mind I would assume Peter has used or developed an algorithm which retains program information when using the EQ and/or volume control. That's actually an interesting topic and something I'd like to dig into if I have time.

Is this true about the foobar EQ (or digital EQs in general)?

Reply #4 –
Kind of, sort of.

Since you're using an outboard DAC, find out if it support 24 bits. If it does, and you output 24 bits to it, this bit depth reduction becomes (even more) meaningless, in regards to volume adjustment (or more correctly, attenuation).

As for equalizing, I have no idea.

Is this true about the foobar EQ (or digital EQs in general)?

Reply #5 –
Lowering the level of a signal will bring it closer to the noise floor in the digital or analogue domain (unless you use floating point calculations in digital processing, but even then you get the same effect from your DAC later down the signal path.)

That's just what attenuation is, there's no getting around it! If you lower the level of a signal by 6dB, it loses one bit, so a 16 bit signal only has 15 bits of useful data. It's not so clear cut when you're using EQ rather than straight attenuation but the principle is the same.

Please don't worry about this just for playback though, you're almost certainly not causing problems by doing this. If it sounds better to compensate for headphone curves then just do it, you're quite probably hearing it more "authentically" than if you didn't compensate.

Is this true about the foobar EQ (or digital EQs in general)?

Reply #6 –
processing is made in a 64-bit floating point engine, so loss as such is nonexistant. Then using 24-bit DAC even so renders that factor uninteresting.

<long unintersting babble about the graphics equalizer being primitive and unprecise and rant about convolver or crying out to people with skills and time to create a decent parametric EQ>

Is this true about the foobar EQ (or digital EQs in general)?

Reply #8 –
From the foobar 'About' dialog:
Quote
SuperEQ library
Copyright &copy; Naoki Shibata, http://shibatch.sourceforge.net/
Distributed under terms of LGPL, modified sources available for download separately.

Is this true about the foobar EQ (or digital EQs in general)?

Reply #9 –
Quote from: tgoose on
Lowering the level of a signal will bring it closer to the noise floor in the digital or analogue domain (unless you use floating point calculations in digital processing, but even then you get the same effect from your DAC later down the signal path.)


True, but please let me clarify. With an analog signal that is attenuated, you lower the noise in the system along with the signal. So you preserve the SNR and the only limit is the noise floor [DR = SNR for an analog system]. In the digital domain, DR and SNR aren't correlated because the noise becomes part of the signal going through an ADC. Therefore, you don't get any benefit from running a 'hotter' signal in digital. I think this may explain the typically poor performance I see in most audio gear using a DSP to do filtering. As the volume lowered I see lots of gear that drastically changes the frequency response of the filter.

Would anyone be interested in seeing the performance of a foobar filter and/or volume control? I have an Audio Precision close to my computer but it only has analog inputs so we'd be limited by the [probably crappy] DACs in my computer. Still, it could be revealing. Anyone want to offer anything they'd like to see tested?
 

Is this true about the foobar EQ (or digital EQs in general)?

Reply #10 –
I think that's a rather fuzzy understanding of analogue and digital.

For one thing, if there is noise inherent in the "capture" stage (let's pick examples: microphone for analogue, A/D for digital) that noise is "part of the signal" going forward in both cases.

If you reduce the energy in a given EQ band later, the lowest noise you can achieve is either the noise in the original capture reduced by the EQ, or (if it's a _very_ steep cut, or a low noise source) the noise floor of the processing and output stage. It's likely the analogue electronics can be made to have less noise than the microphone. It's likely the digital processing can be made to have less noise than the A/D (since arbitrary bitdepths are available for processing).

Then you have the output stage - either purely analogue electronics, or a DAC. With 24-bits, it's the analogue electronics of the DAC that set the noise floor, rather than the digital data itself. With a 16-bit DAC, then yes, you'll have to re-introduce noise or distortion to get down to 16-bits from the arbitrarily high bitdepth you used for processing - but we're still only talking about 80-100dB below full scale.


In both domains, it's possible to perform the EQ processing using a bad circuit or bad maths (insufficient resolution - typically IIR filters which audibly shape their own quantisation noise) - or audibly "bad" algorithms (either domain).


Audio Precision, though a nice piece of kit and well worth it for measuring analogue outputs (even from digital gear), is largely easily and freely replicated in the pure digital domain without ever sending the signal through your sound card - if you're just interested in the digital processing.

Cheers,
David.