If Kodi’s internal volume control is lowered then the dynamic range is effectively compressed since it is manipulating the LPCM stream itself. In a setup such as your running the optimal configuration would be for the volume in Kodi to be at max, and to not use the volume control on the OSMC remote (or in Kodi) at all. I didn’t know if this was how you were running or not, so I asked.
Ok, understand.
The volume on the Vero is typically set towards the upper end of the scale. At least more than 80%. The round dial is set to approx 11 o’clock.
Why? It should be just scaled down ie all values reduced by XdB. Or does Kodi have some sort of ‘loudness’ adjustment?
I don’t know. I do know if I play “So What” from Miles Davis’s Kind of Blue album with Kodi at 50% volume with my AVR turned up to compensate it sounds quite different than with Kodi at full volume and the AVR turned down so my speakers are producing a similar SPL.
Woah! That’s news. I wonder if Kodi’s softvol does indeed have some compression and maybe on Windoze doesn’t use the softvol but controls the main soundcard volume instead.
OK, did the things you asked me to do. I attached my Focusrite USB audio interface to both my win10 3.5mm audio port and to the Vero4K+ 3.5mm audio port and recorded directly from the output (rather then using speakers and record back with a good condenser mike what they play). The vero sound was turned almost fully up (11 o’clock on the volume dial)
However, the result is EXACTLY THE SAME with direct recording as using speakers…the Vero compresses the signal. To explain once more what I did: I selected a very specific section of 2 seconds (without voice over, so soft background music only), and amplified the whole clip until I get the same level for that 2 sec part (-27db in this first case). Then I measured the peaks of the voice over: Windows computer -5db, Vero -17db… the vero DEFINITELY COMPRESSES peaks much more… Chart below is one channel only, but both channels show same thing.
Then I also did the same test with the Vero volume reduced to 25% (3 o’clock). Obviously I then had to amplify more to get the reference part at -27db, but the peaks still remain at -17db. So the Vero does not compress more at lower gain levels than at high levels (as Darwindesign suggested).
Then, I re-did the same test (Windows vs Vero) with another movie from another person. Exactly same results: the vero compresses much more:
So, I do not know what else I can try… for me it’s clear: the audio section of the/my VERO is not on par with a Windows machine, or a 15 old Medi8er mediaplayer? And I have proven before that using the Toslink output, I had exactly the same compression.
Hope you specialists can find a parameter I can tweak (but pls note I’m a video guy, not a Linux wizzard… I managed a unix CAD/CAM machine 35 years ago, but remember nothing of that).
Is the clip you used for that the same (or part of) the clip you sent us? I’ll try to do exactly what you did. Where is the 2 seconds in the timeline?
@darwindesign I couldn’t tell any difference between Kodi low vol + AVR high vol and Kodi high vol + AVR low vol, but it’s tricky to get a true A/B on that. Tested with LeoB’s clip and some random Miles Davis.
So you tested it at roughly 90 and 25% output from the Vero but didn’t test it at 100%?
Could you post logs for us to check your setup?
Interesting. The reason why I mentioned that track was that it was rather dynamic and on my system the cymbals seem to diminish much faster than the horn. As I shifted Kodi volume down more the clarity, distinctiveness, and sparkle (for lack of a better word) of the entire thing seemed to decrease rather noticeably. But indeed, as I lack the ability to scientifically test this I am left only with my subjective perception and biases that are inherent in that.
I don’t think I have that very track. I’m prepared to believe that Kodi’s softvol is not the best but I need more evidence for an actual dynamic range compression. I’ll knock up some sox clips at set dB intervals and see what that does.
Indeed, approx 25% and 90%. No difference whatsoever. Seen that, I cannot imagine that between 90% and 100% all of a sudden things will drastically change.
I remember from working with (professional) analog audio mixers in the past, to always shoot for approx 70%. The people who taught me said never to use max positions. That still fuels my internal resistance to go over 90%.
Yes, movie A (first image) is still that same clip. Movie B is a completely different movie (from a different editor), but shows same patterns.
I’ll check this weekend how to make & upload logs…
But in this case 100% means that the audio isn’t being modified and also this volume control in Kodi isn’t analog.
1 Like
My understanding is that applies when you are recording live - comes from the old days when things got rather non-linear the closer you were to 0dB on the tape. With digital, it’s still a good idea because you don’t want risk the very hard clipping.
But if you are re-playing a digital stream that can’t go over 0dB there’s nothing against turning it up to max with the benefit you are not asking software to scale the signal (which I think Kodi does without dithering but don’t quote me).
Thanks. I realised that when I re-read your post. But here’s the thing - If I capture the clip from ‘Stereo Mix’ on my PC and normalise it so the peak around the 28s mark is 0dB, the bit you adjusted to -27dB is reported as -32dB by Audacity. So I can’t amplify it by 5dB as you did.
There seems to be some compression going on even in that first capture you did.
Update: I added ffmpeg to my Audacity and imported the AAC direct. Same result - -32dB in those 2 seconds.
Your analogue instincts are wrong in a digital world. Digital volume controls should always be kept at 100% unless reducing them is your only option. The volume control should be an analogue process, if possible. If you reduce the signal digitally then, at best, you are reducing the bit-depth of the samples before they are sent to the DAC, so you’ve switched from (say) a 16-bit signal to a 14-bit one or lower; at worst, the resampling may not even be done correctly. You want the Vero to output the digital signal completely untouched, and then have the amplifier control the volume - in the analogue domain, if possible.
Try it with the volume at 100% and see what happens. There may very well be a difference between “altered” and “not altered” which only becomes apparent with the volume at exactly 100.
[EDIT: Okay, there is an exception to the above guideline: if the DACs have a much higher resolution than the signal ever can - for example if you never decode a signal with more than 24 bits, but your device has high-quality 32-bit DACs - then you can get away with using a digital volume control. In that case, full volume output involves multiplying each sample number by 256 before it reaches the DACs, and you reduce the output volume by using a smaller multiplier; so long as the multiplier is at least 1, you don’t lose any resolution from the original signal. But only a fairly high-end device will have decent 32-bit DACs; something in the Vero 4K’s price range is unlikely to work like that. So, a device like the Vero should always have its volume set to 100.]
3 Likes
Last test: Vero volume cracked all the way up to MAX (and normalized the same reference part to -27 db as all other tests). Unfortunately exactly the same audio compression is seen as in any other test I did with the Vero.
Log created as requested: https://paste.osmc.tv/rozahibufu
I updated the Vero just before making the log… it was on the March22 version, now on Oct22 version.
I tried something else tonight… I had a cheap-ass USB soundcard laying around (an Ali thing for 3 euro). Did the test on the Vero, AND used the same usb soundcard on my win10 machine. Using it on the Vero gives me that compressed signal as ever before, using that exact same soundcard on my computer gives me the dynamics I got on the computer build-in audio or the old Medi8er mediaplayer… so it really seems to be the Vero which is compressing…
1 Like
Try turning off ‘Maintain original volume on downmix’. The description talks about compressing the audio. I’ve no idea why this setting is needed or why it would affect stereo but by default it’s off.
And we always recommend turning off Sync playback to display.
1 Like