Broken VU output amplitude after update

Do you understand the profound impact of the DRC here...?
everything is distorted...
;-)
No. Sorry. I have a hard time with understanding what you write.

I know what DRC is but cannot see the relevance for measuring the dBFS.
 
No. Sorry. I have a hard time with understanding what you write.

I know what DRC is but cannot see the relevance for measuring the dBFS.
Look for the discussions from a few months ago about this DRC... which were swept under the rug but which render your rigorous considerations regarding the VU meter obsolete... and which actually had a heavy impact in many sensitive subjects here .. never addressed... a very clumsy Wiim workaround...
 
measure for vumetre of ultra before or after drc? ;-)
 
Last edited:
Look for the discussions from a few months ago about this DRC... which were swept under the rug but which render your rigorous considerations regarding the VU meter obsolete... and which actually had a heavy impact in many sensitive subjects here .. never addressed... a very clumsy Wiim workaround...
Are you talking about the limiting that WiiM use to prevent clipping? Yes I can see that would complicate things.
 
Before, of course.
Do the compression even happen on the digital outputs? Or only before the DAC?
I'll let you investigate... ;-)
But realizing the impact on the rest of the circumstances used on different Wiim systems (not only ultra -vumetre) remains quite essential, it seems to me...
;-)
 
I'll let you investigate... ;-)
But realizing the impact on the rest of the circumstances used on different Wiim systems (not only ultra -vumetre) remains quite essential, it seems to me...
;-)
So you think WiiM is using DRC in other cases than to avoid clipping in the DAC after applying gain in the EQ settings?

We need to have that confirmed by the @WiiM Team .
 
No. Sorry. I have a hard time with understanding what you write.

I know what DRC is but cannot see the relevance for measuring the dBFS.
It is documented here
 
It is documented here
If the WiiM algoritme always reduce any applied gain then we don't need a clipping indicator but it would still be good to have an real indication of how close the signal level is to 0 dBFS.

On input it will show the dynamic range of the source and on output we could see if this "advanced algorithm" is about to getting applied.
 
I just did a few VU-meter tests myself:
  1. VU meter "Original Signal" setting
    • With -14dBFS RMS / -11dBFS peak 1kHz sine the VU meter shows about -11dB
    • With -14dBFS RMS / -8.5dBFS peak pink noise the VU meter shows about -6dB
    • With either input signal VU meter level doesn't change when I enable EQ or RoomFit with +6dB boost at 1kHz, nor when I added +7dB of pre-gain boost to the input, or changed output level.
  2. VU meter "Output Amplitude" setting
    • Pink noise and 1kHz tone with the same -14dBFS RMS level result in different VU meter readings (i.e. like above).
    • VU meter level is affected by input pre-gain, output level, EQ and RoomFit.
My conclusion based on above results is that the current implementation of VU meter doesn't match signal peak nor signal RMS levels.

Regarding DRC, IMHO @canard is implying that the story is more complicated than it really is.
According to my tests, WiiM only apply DRC/limiting when the output of the device would be pushed over 0dBFS in digital domain, to avoid hard clipping.

DRC seems to be triggered only when output volume is at (or close to) 100%, combined with some kind of digital boost (either pre-gain, EQ or RoomFit) and when playing back content with peaks close to 0dBFS. I.e. even with boosts in the flow DRC won't be triggered if volume isn't high enough to end up above 0dBFS at DA conversion (or digital output).

For sure DRC/limiting is much more gentle to the ear compared to digital clipping, so I personally find this limiter to be a smart safeguard, useful for casual users who might not know how to achieve optimal gain staging.
 
I just did a few VU-meter tests myself:
  1. VU meter "Original Signal" setting
    • With -14dBFS RMS / -11dBFS peak 1kHz sine the VU meter shows about -11dB
    • With -14dBFS RMS / -8.5dBFS peak pink noise the VU meter shows about -6dB
    • With either input signal VU meter level doesn't change when I enable EQ or RoomFit with +6dB boost at 1kHz, nor when I added +7dB of pre-gain boost to the input, or changed output level.
  2. VU meter "Output Amplitude" setting
    • Pink noise and 1kHz tone with the same -14dBFS RMS level result in different VU meter readings (i.e. like above).
    • VU meter level is affected by input pre-gain, output level, EQ and RoomFit.
My conclusion based on above results is that the current implementation of VU meter doesn't match signal peak nor signal RMS levels.

Regarding DRC, IMHO @canard is implying that the story is more complicated than it really is.
According to my tests, WiiM only apply DRC/limiting when the output of the device would be pushed over 0dBFS in digital domain, to avoid hard clipping.

DRC seems to be triggered only when output volume is at (or close to) 100%, combined with some kind of digital boost (either pre-gain, EQ or RoomFit) and when playing back content with peaks close to 0dBFS. I.e. even with boosts in the flow DRC won't be triggered if volume isn't high enough to end up above 0dBFS at DA conversion (or digital output).

For sure DRC/limiting is much more gentle to the ear compared to digital clipping, so I personally find this limiter to be a smart safeguard, useful for casual users who might not know how to achieve optimal gain staging.
Indeed, it's more complex... but my assessment is slightly more negative...because it systematically hides many negative situations where positive gains in corrections, etc., will be unreadable due to saturation, where a simple "overload" in Wiim Home (like streaming white noise at 0dB would allow you to adjust the gains accordingly...)

It's a very relative camouflage...because in action, without digital clipping, it's much more than unpleasant...it creates significant distortion
. ;-)

it makes many regrettable situations illegible
Basically, digital clipping has the advantage of being very clear and explicit.
 
Last edited:
Indeed, it's more complex... but my assessment is slightly more negative...because it systematically hides many negative situations where positive gains in corrections, etc., will be unreadable due to saturation, where a simple "overload" in Wiim Home (like streaming white noise at 0dB would allow you to adjust the gains accordingly...)

It's a very relative camouflage...because in action, without digital clipping, it's much more than unpleasant...it creates significant distortion
. ;-)

it makes many regrettable situations illegible
I'm with you that optimizing the signal level in WiiM devices is not intuitive with the current implementation.

I just don't think that is a very big deal in most cases.
 
I just did a few VU-meter tests myself:
  1. VU meter "Original Signal" setting
    • With -14dBFS RMS / -11dBFS peak 1kHz sine the VU meter shows about -11dB
So WiiM is trying to show a dBFS peak value on a dB meter. Don't make sense.
    • With -14dBFS RMS / -8.5dBFS peak pink noise the VU meter shows about -6dB
    • With either input signal VU meter level doesn't change when I enable EQ or RoomFit with +6dB boost at 1kHz, nor when I added +7dB of pre-gain boost to the input, or changed output level.
On the input there is no gain applied. That is fine. Can be used to see the dynamic range of the original signal if it really does show the absolute dBFS peak value.
  1. VU meter "Output Amplitude" setting
    • Pink noise and 1kHz tone with the same -14dBFS RMS level result in different VU meter readings (i.e. like above).
    • VU meter level is affected by input pre-gain, output level, EQ and RoomFit.
This should show the resulting peak dBFS value after applying all types of gain and volume control. This can then be used to check how close the signal is to digital clipping.
My conclusion based on above results is that the current implementation of VU meter doesn't match signal peak nor signal RMS levels.
I don't see how you can use RMS for a digital playback. Maybe I am missing something here?
For sure DRC/limiting is much more gentle to the ear compared to digital clipping, so I personally find this limiter to be a smart safeguard, useful for casual users who might not know how to achieve optimal gain staging.
It would however be good to know when this happens, as it will probably change the sound profile. If you use the WiiM at fixed (100%) volume this could easily impact the signal forwarded to an external DAC/Amp with own volume control.
 
A
I'm with you that optimizing the signal level in WiiM devices is not intuitive with the current implementation.

I just don't think that is a very big deal in most cases.
from the moment when the Wiim devices are likely to modify the signal, correction etc (without automatic gain adaptation)...I don't really agree with you
;-)
 
What do you mean? The max dBFS value is defined to be 0. There are no positive dBFS values.
You are correct that for transport e.g. by optical SP/DIF or SMTPE the number range is limited to 0dBFS. And anything above 0dBFS of the DAC will cause distortion. Fully agreed.

But of course there are positive dBFS values internaly in digital audio signal processing.
The reason is that for internal digital processing typically number formats with higher dynamic range, i.e. more bits are used than the ADC and DAC have.
I think this is the point where our discussion is centered.

Let's take your example of an ADC or DAC having 24 bit. Then their number range is limited to 2^24-1 integer or 0xffffff Hex. Let us define this as reference for full scale, 0dBFS. ADC and DAC output signal number range define full scale 0dBFS. I think we agree so far.

To apply EQ with positive gain, which will increase signal amplitude beyond 2^24-1, or effects in digital audio like reverb etc., internally to a digital signal processing system like a DAW or the WiiM a larger number format is used, tpically. For example 32bit integer with 2^32 number range 0xffffffff, or 32bit floating point numbers. In Steinberg Wavelab, the leading digital audio software for audio mastering, 64bit float can be used.

With 8 bit more for internal processing at 32bit, you gain roughly 48dB internal dynamic range. Therefore, several 10dB gain above the full scale of the ADC or DAC can be applied, giving internal levels of +30dBFS, for example.

Of course, for final DAC to analog voltage, the signal must be limited to 24bit 0dBFS, otherwise you will get distortion. This can happen easily if the digital gain of EQ or effects is too high. Therefor, a final volume control is required, with negative gain, to keep the highest peak signal just below the number range or 0dBFS of the DAC. In music produstion this is assured by dynamic compressors, limiters or more sophisticated loudness maximizers which give a brick-wall signal peaking constantly at 0dBFS. But before, signal amplitudes above 0dBFS must be used that the limiter has something to work on. This process is monitored in a DAW by a True Peak Meter with a scale going well above 0dBFS to see if there are problems with overdriving the DAC.

Example screen shot with a 100% FS 24bit sine, and no gain applied. True Peak Meters are showing 0dB(FS).

1757884631057.png

Now the same with 6dB gain applied. True Peak Meters are showing ... +6dB. No internal distortion because of the 64 bit number format. But of course distortion in the output, either analog (limited by the DAC bit range) or digital (limited by the protocol or data format bit range).

1757884766644.png

To prevent overdriving the output DAC of the WiiM by too much internal pre-gain, EQ gain, RoomFit gain, a proper True Peak Meter right after the final volume control (unfortunalty in %, with -10% seems to be -6dB gain) and before the DAC is desirable. To see if it peaks above 0dBFS, and to adjust volume that is does not.
 
0 dB is mapped to -18 dBFS
The mapping of dBFS, defined on the digital side by the DAC, to analog voltage in whatever analog dB definition is more or less arbitrary by electronic analog gain. 0 dB what analog? In #9 -18dBFS would be 0dBVU or +4dBu, which is actually 1.228 Volts RMS.

On the WiiM you can adjust for 0dBFS on the digital side different line out voltages on the analog output side:
0dBFS = 2V RMS = +4.2dBVU
0dBFS = 1V RMS = -1.8dBVU
and so on.
However, this is completely irrelevant for discussion of internal digital number formats.
 
Back
Top