Print

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

[6233638]

Would it be possible to unlock the position of the Volume Levelling DSP, or have Process independently of internal volume bypass it when enabled for VST plugins?
It seems that it must only apply to the actual volume control (which I have disabled) and not any other gain adjustments that are made in the audio path.

The plugin I'm trying to get working uses absolute values (e.g. -2dB) and with Volume Levelling enabled, those values change with every album being played. -2dB becomes -16dB, if the album gain is -14dB for example.

[Matt]

Volume Leveling happens first because it's tied to a file.

Other DSPs happen just-in-time.  At that point there's no way to know what file the data is from (it could be a mixture during a fade, etc.) so that method won't work for a file-specific change like Volume Leveling.

[6233638]

Is there any way to "bypass" that for a VST plugin which has Process independently of internal volume enabled?

Have Volume Levelling work as it normally does at the top of the stack, but for the input to a VST plugin with that option enabled, boost the gain by the inverse of the current adjustment (+Total Volume Change) and then apply the reduction again to the VST output?

That sounds really awkward, but some plugins just won't work unless they have the original volume.

To make it clearer: after a couple of podcasts and videos that were very uncomfortable to listen to due to audio clipping, I've been demoing some plugins that are supposed to help "fix" audio clipping - or at least make it less grating.
But for them to work correctly, they need to be set to only act on audio within a certain range, such as -0.2dB - sometimes audio is clipped when it was recorded, and then reduced slightly when it was encoded.

[mojave]

Do you still want to use volume leveling on these podcasts and videos? If not, you could create a separate zone and use ZoneSwitch to switch to the zone with the plugin for podcasts and videos. If this doesn't work for you, I think you can use Convert Format (at least for the audio files) to convert them to the same format but with Volume Leveling already applied. Now you can play the converted files through the plugin without having to use Volume Leveling in real time.

What plugin are you trying? There may be a way to replicate it in JRiver's PEQ. For example you might try the limiter and add a brick wall limiter to cut the peaks off in a better sounding manner.

[6233638]

Do you still want to use volume leveling on these podcasts and videos? If not, you could create a separate zone and use ZoneSwitch to switch to the zone with the plugin for podcasts and videos. If this doesn't work for you, I think you can use Convert Format (at least for the audio files) to convert them to the same format but with Volume Leveling already applied. Now you can play the converted files through the plugin without having to use Volume Leveling in real time.

I should have said that it's not just podcasts and video I'm using it on. If the clipping threshold is only set to around -0.2dB it seems to be non-destructive when playing music, and can help out with highly compressed tracks as well.

I would much rather use a plugin for this than edit it into my music files - it may not always be good to leave it enabled, and similar to keeping RAW files for photography, while the results are good today, who knows what may be available in five or ten years time.

What plugin are you trying? There may be a way to replicate it in JRiver's PEQ. For example you might try the limiter and add a brick wall limiter to cut the peaks off in a better sounding manner.

I've been trying out a few different plugins that attempt to recreate the original audio rather than simply tone down the clipping artifacts, and they can do a surprisingly good job.

Actually, I think Perfect Declipper is one of the plugins that doesn't require the clipping to be at or near 0dB, but there are some really good samples on their site: http://www.perfectdeclipper.com/samples/

[Matt]

The topic of declipping piqued my interest, so I did a little research.

A common approach is to declip a PCM signal in the time-domain using spline interpolation when the signal flat-lines.  This is a neat approach because it's perfectly transparent on a signal that doesn't flat-line.  However, this doesn't really help "soft knee" limiting that's common in mastered CD audio.  For example, I don't think one of these would engage for something like Metallica - The Day That Never Comes, which is a track listed as an example of clipping on one of the declipper sites.

The other approach is to switch to the frequency domain, and try to rebuild regions that are too hot using interpolation from surrounding frequency blocks.  This "reconstructs" the signal.  However, it's a heavier filter that would do more harm to the signal when there's no clipping.  Also, from my listening to the declipper samples, I didn't really like the results much (loudness pulses, etc.).

[6233638]

The other approach is to switch to the frequency domain, and try to rebuild regions that are too hot using interpolation from surrounding frequency blocks.  This "reconstructs" the signal.  However, it's a heavier filter that would do more harm to the signal when there's no clipping.  Also, from my listening to the declipper samples, I didn't really like the results much (loudness pulses, etc.).

I actually don't like the results I got from Perfect Declipper with music - but if you look at the "extreme clipping" sample, it shows a significant improvement. (and I suspect most of their samples are "turned up to 11" to make the difference obvious)
I still wouldn't listen to music that had been clipped to that degree, but if a podcast or a recording of a live show that I want to listen to is clipping badly - which is what got me looking into this - a declipper can make the difference between the audio being grating to listen to, and relatively OK.

I downloaded the trial of iZotope RX2 today (a friend uses it in video production and recommended it) which gives you the option to set a hard threshold (I found -0.3dB to work best) rather than trying to analyze the music, and while it may not be as good as a well mastered source, if your only option is a clipped CD, it seems to make the difference between it being fatiguing four tracks into the album, and being listenable all the way through.

Because you can set it to only work on samples that are -0.3dB or louder, it won't touch well mastered music at all (unlike Perfect Declipper) and typically if something is that loud, it's probably clipping - or just about to.

It's a far more subtle/cautious approach than Perfect Declipper that maybe won't give you as good results if you've spent a lot of time tuning it for a specific track, but it seems to work well with everything I've thrown at it so far, and doesn't interfere with properly mastered tracks - which is important to me. I want something I can just leave running without thinking about it rather than having to play around with controls all the time. (Perfect Declipper has far too many variables and not enough documentation)

From what you describe, it seems to use the first approach, or a hybrid approach that primarily only touches samples that have been flattened. If you set the threshold too low, it will start to affect other samples and sound terrible.
But because you have to set that hard limit at (for example) -0.3dB, you won't be able to use that if ReplayGain is working.

[Matt]

If you have a sample clip where iZotope is helping a lot, I'd like a copy (matt at jriver).

I'm interested in if a declipper working in the time-domain would be useful as a core component of the JRiver audio engine.