Print

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

[lendall]

I have been using "Apply DSP Effects" when I run Sync Handheld because I listen to music on my mobile in noisy places and I want the level to be compressed.  Therefore I select both Volume Leveling and Adaptive Volume - Night mode, the latter for purposes of compressing the level within each track.  In listening to music on my phone using the Neutron app, I have experienced some radical issues of volume shifts and distortion and, since both the phone and the app are new, I was blaming them.  However today I started to investigate further by looking first at the original FLAC file in my MC music library and then at the lower bandwidth OGG file cached by MC for use in running Sync Handheld.  I opened the original FLAC in Audacity and found the first of the two images attached. Then I opened the cached OGG file (to which the DSP effects had been applied as well as the compressed bandwidth and change in file type) and found the second.   Based on listening to tracks on my phone, I have the feeling that there are enough of these anomalies that the only practical solution is to wipe out my entire Sync Handheld conversion cache and start over.

But what is going to prevent this problem for happening again?  The original FLAC file may be slightly hot, but what could possibly explain MC doing what it did -- i.e., making the entire track radically clipped and distorted -- in making the conversion?   ?

[blgentry]

To be sure about this, I'd need to know how much compression Night Mode applies.  I'd also need to do some tests of my own.  That being said, what you've posted might be TOTALLY normal.  Here's why:  You're looking at a linear view of the waveform.  As you probably know, hearing (and most other human senses) are logarithmic.  This is why the deciBel scale was invented; because it's a logarithmic representation of magnitude (Volume).

If you change the view in Audacity to Waveform dB, instead of Waveform, you'll see a more realistic version of what's going on with the files.  Click the down arrow in the upper left corner of the track and select it.

I'm *guessing* that JRiver's night mode applies something like 10 to 15 dB of compression with a medium low threshold like -30 dB.  If I'm right, you're going to see something like 12 dB of extra gain throughout most of the track, but the peaks will be unchanged, or affected only slightly.

Further, I don't see any clipping in your waveforms.  Audacity normally indicates clipping with red indicators to make them really obvious.  I understand you saying it looks very different and you probably just mean "turned way up" for clipping.  But that's an incorrect use of that term.  I point this out because terminology is important to me and for good communication in general.  I've also talked to LOTS of people who use the term clipping to mean all kinds of different stuff that it doesn't mean.  Anyway, just know that the waveform doesn't appear to be clipped.  

In short, I think MC is doing it's job exactly as it's supposed to.  But someone else here is bound to know more about this than me.

Brian.

[mattkhan]

"show clipping" is off by default in audacity (http://manual.audacityteam.org/o/man/audacity_waveform.html)

Switching to Waveform dB will just make it look like there are higher levels overall as relatively more of the vertical axis will be given to low level content. If anything, the pure waveform view makes it easier to see clipping and that 2nd graph looks pretty heavily clipped to me.

[mattkhan]

I tried to recreate this using some random track from my library. 2 comments

1) where is the option for "apply DSP" via sync handheld? I don't see that, volume levelling yes but apply dsp no
2) convert format does have this option and this results in clipping in the output whenever I turn on night mode irrespective of whether volume levelling is on, this applied to wav & ogg output formats so I would assume this is independent of the output encoder

[blgentry]

1) where is the option for "apply DSP" via sync handheld? I don't see that, volume levelling yes but apply dsp no

Tools > Options > Handheld > Conversion > Apply DSP .  Check the box to on.  Then below that, click DSP Settings.

Brian.

[mattkhan]

Tools > Options > Handheld > Conversion > Apply DSP .  Check the box to on.  Then below that, click DSP Settings.

ok so I just stumbled across http://yabb.jriver.com/interact/index.php?topic=72196.0

this seems a rather odd limitation albeit a tangent for this thread (as it must be the same convert format functionality used by handheld sync) so the same potential clipping issue applies. FWIW the clips are v short as far as I can see, it's not flatlining against 0dBFS in the case I tried, it's perhaps more of an intersample overage issue.

To the OP, can you post the track you're using somewhere?

[lendall]

Thanks to everyone for their comments.  Attached is the same segment of audio with "show clipping" turned on in Audacity.  Looks pretty darned clipped to me.  It was also interesting that Mattkhan got the same results I did using apply Night Mode, with or with Volume Leveling, and irrespective of the conversion file type being used.  Is it possible that we are the first MC users ever to try this?  Does Night Mode induce severe clipping when you are just using it for playback (not in file conversion)? 

Perhaps I should have mentioned that the track I posted for this sample had been run through analysis in MC.  I have had a few anomalous (meaning in this case inconsistent) results with Analyze Audio recently -- a track analyzed one day will get a different R128 number the next.  Not all tracks, only some.  I am going to go back and check the original FLAC track and run Analyze Audio on it again to see I get different numbers.  Presumably Night Mode and Volume Leveling are both using the settings derived from running Analyze Audio.  If not, how would they work?  I mean, one of the primary purposes of using Analyze Audio to get the R128 number, etc., is so that features like Volume Leveling and Night  Mode will work.  Meanwhile I would be interested to know if the file Mattkhan used and on which he also found clipping when he applied Night Mode had also been analyzed. 

[mattkhan]

My files are all analysed.

One thing I noticed is that the "clip" is v short so I am unsure whether it really is hard clipping vs just having a full scale signal in there. I didn't get time to examine further but you could try zooming in to examine that.

[lendall]

Here I have zoomed in on a section with what still appears to me to be severe clipping.  
What is interesting is that this OGG track in my conversion cache was created using both Volume Leveling and Adaptive Volume.  The Volume Leveling screen has an option for "Clip Protection," which I have selected (pull down menu). "Clip Protection" should prevent a track from reaching a level where it clips, right? I believe that the instructions suggest that if one activates Volume Leveling one may also wish to use Adaptive Volume, so MC is endorsing the idea of using both together.  

[wer]

This is a somewhat old thread, but I am seeing a similar issue so I thought I would post here.
(Note: I only use night mode with video)

I agree that Night Mode seems to be radically misbehaving.  From my own analysis using JRiver's analyzer view, night mode is boosting the volume on everything, including the peaks.

In the nightclub scene from The Dark Knight, with night mode off the peak levels are at around -14.  With night mode on, the peaks reach -4.

This is exactly the opposite of what should be happened with peaks. Low passages should be boosted, peaks should be reduced.

Night Mode is boosting low passages AND peaks, resulting in an overall dramatic increase in average volume. 

It's unusable in current form for me, but is anyone else seeing this, if they even ever use night mode?