Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[BCZ]

I wanted to take a look at what the loudness feature is doing on JRiver, so I fired up the analyzer in DSP Studio to take a look.

It seems like the volume control is wrong when loudness is activated. For example, when I increased the volume by 10dB, I see the 1-4KHz band (which are not affected by loudness) go up by around 15dB, but the bass band went up 10dB.

I don't think that's the right way to implement loudness. The loudness contour is referenced at 1KHz. Therefore, a 10dB increase on the volume should mean a 10dB change on the 1-4KHz band, and the bass should increase only say 7dB instead of 10dB.

Also, on a side note, I feel the loudness contour is too aggressive. Is there any way to adjust it?

[~OHM~]

have you seen this...

https://wiki.jriver.com/index.php/Volume

[stewart_pk]

The article above discusses it and in reference to calibration if you want to do that.
The setting is: Options > Audio > Volume > Internal volume reference level.

[BCZ]

I have.

Either way, the change in volume with loudness on is wrong. I just had a friend, who did not know my speaker had loudness on, and he told me my speakers sounded weird as the volume went up as it felt the 200-5000Hz went up a lot more than the rest. He pretty much nailed the range without knowing anything about JRiver's loudness implementation, or even know that it was applied.

If I turn up the volume by 10dB, I expect the volume to go up by 10dB (or less if loudness is on), not go up by 15dB.

[RD James]

Your speakers need to be calibrated to an 83dB reference level and all volume control handled by Media Center for this feature to work correctly.
If you don't have a calibrated volume level, I recommend that you do not enable the Loudness feature.

[BCZ]

Your speakers need to be calibrated to an 83dB reference level and all volume control handled by Media Center for this feature to work correctly.
If you don't have a calibrated volume level, I recommend that you do not enable the Loudness feature.

I did do that.

Even if I didn't, it shouldn't matter. The equal loudness contour is supposed to be a large boost in the lower frequencies and a smaller boost in the high frequencies at lower levels. Therefore, as you increase the volume by X, the midrange increases by X, but the lower and high frequencies increases by an amount less than X.

What JRiver does is if you increase the volume by X, the lowest of the lower frequencies increases by X, but the midrange increases by an amount greater than X. While this method can achieve the same sonic experience as the above, the SPL on the volume slider in JRiver does not match the output SPL.

Again, if you don't believe me, turn on loudness, go to DSP studio's analyzer, throw a pink noise on, then increase the volume by 10dB on the JRiver's volume slider. You'll see on the analyzer that the midband is clearly increased by more than 10dB.

[RD James]

I did do that.
Even if I didn't, it shouldn't matter. The equal loudness contour is supposed to be a large boost in the lower frequencies and a smaller boost in the high frequencies at lower levels. Therefore, as you increase the volume by X, the midrange increases by X, but the lower and high frequencies increases by an amount less than X.
What JRiver does is if you increase the volume by X, the lowest of the lower frequencies increases by X, but the midrange increases by an amount greater than X. While this method can achieve the same sonic experience as the above, the SPL on the volume slider in JRiver does not match the output SPL.
Again, if you don't believe me, turn on loudness, go to DSP studio's analyzer, throw a pink noise on, then increase the volume by 10dB on the JRiver's volume slider. You'll see on the analyzer that the midband is clearly increased by more than 10dB.

Loudness takes effect when you reduce the volume, not when you increase it. It is to compensate for the way our hearing responds differently at low volume levels.
I'm sure that it could be made to reduce the high and low frequency response when you increase the volume above the reference level, but that's not how it currently operates. Above the reference level, you just get standard volume adjustments.

I have now tested using Voxengo Span, which is a free VST plug-in that has additional features compared to the built-in analyzer.
Smoothing and averaging a pink noise signal over 30 seconds shows that volume changes above the reference level do not affect the frequency response, and that all increases in volume match Media Center's volume slider exactly.
 
Please note that if you are listening on a calibrated playback system at +10dB as you claim, you should be limiting playback to <1 hour a day or you risk permanent hearing damage.

[BCZ]

Loudness takes effect when you reduce the volume, not when you increase it. It is to compensate for the way our hearing responds differently at low volume levels.
I'm sure that it could be made to reduce the high and low frequency response when you increase the volume above the reference level, but that's not how it currently operates. Above the reference level, you just get standard volume adjustments.

I have now tested using Voxengo Span, which is a free VST plug-in that has additional features compared to the built-in analyzer.
Smoothing and averaging a pink noise signal over 30 seconds shows that volume changes above the reference level do not affect the frequency response, and that all increases in volume match Media Center's volume slider exactly.
 
Please note that if you are listening on a calibrated playback system at +10dB as you claim, you should be limiting playback to <1 hour a day or you risk permanent hearing damage.

Ahhh I see where the misunderstanding is. I was talking about going from say -30dB to -20dB. It would be terrible to listen above reference volume a lot. Although I did wish there were changes when going above reference. Namely the curve would look more and more like a "n" shape. But I digress.

I also now see where you're coming from. You're correct that the loudness contour is meant for lowering volumes, and the way this is implemented reflects this, which is sensible. However since most people will be listen well below reference almost all the time, it may be confusing to have the volume slider work the current way.

[flac.rules]

I agree with BCZ, the equal loudness contours are normalized compared to the 1k-level, the most "natural" behaviour would be that 10 dB lowering of volume turns it down 10 dB at 1k.

[RD James]

I agree with BCZ, the equal loudness contours are normalized compared to the 1k-level, the most "natural" behaviour would be that 10 dB lowering of volume turns it down 10 dB at 1k.

That's what the current behavior does.
I ran another test with pink noise and the volume level set at -30 dB. With the test file I was using, at 1kHz, that put the pink noise level at -70.2 dB.
With the volume set to -20 dB, the pink noise was at -60.3 dB, and with it set to -40 dB the pink noise was at -80.0 dB.
There's a very slight error there, but I wouldn't be too concerned about an 0.3dB error over a 20dB adjustment. Of course I would prefer that it wasn't there at all.
 
If I was going to complain about anything, it would be that the approximation of the equal-loudness contour is more vague than I'd like.

[flac.rules]

That's what the current behavior does.
I ran another test with pink noise and the volume level set at -30 dB. With the test file I was using, at 1kHz, that put the pink noise level at -70.2 dB.
With the volume set to -20 dB, the pink noise was at -60.3 dB, and with it set to -40 dB the pink noise was at -80.0 dB.
There's a very slight error there, but I wouldn't be too concerned about an 0.3dB error over a 20dB adjustment. Of course I would prefer that it wasn't there at all.
 
If I was going to complain about anything, it would be that the approximation of the equal-loudness contour is more vague than I'd like.

It doesn't sound like it, unless i misunderstand you. If the total level of the pink noise moves down 10 dB, the 1khz wont move down 10 dB, and with loudness applied, that is what corresponds best to what your ears "expect"

[RD James]

It doesn't sound like it, unless i misunderstand you. If the total level of the pink noise moves down 10 dB, the 1khz wont move down 10 dB, and with loudness applied, that is what corresponds best to what your ears "expect"

As I said, I was measuring the level at 1kHz.

[flac.rules]

As I said, I was measuring the level at 1kHz.

Ok, then i misunderstood you, I thought you meant the whole signal when you referred to the level of the pink noise.