Topic: Recorded new show with different volume levels- can't amplify in Audacity

[nicolaidenmark]

Hi,


I recorded a show using my PCM-M10 (a revelation after having used Sony MD and ECM 719 mic for years). But I have a problem: the show was recorded at different volume levels and when I try to amplify the low volume part, Audacity won't let me do it, unless I'll allow clipping, which I won't.

What can do?

I recorded the show in 32 bit floating 96KHZ (max quality available with PCM-M10). Thank you.

My recording clipped during the recording of the show as I had accidentally sat mic sens to high. The last third of the show is without clipping, though. I have normalized the track to. I think. -3 DB.

[nicolaidenmark]

Screen shot added.

[bombdiggity]

Ouch.  It certainly appears you did brickwall some section of it there.  It appears you then reduced the volume of the overall file (I'm assuming that's the normalization you refer to)?

Amplification will apply the same gain to the entire file (or a selected area if you select less than the entire file).  If you're already at or over 0 dB on some part you try to amplify then amplification will just push that part that much further into clipping.  Simple math (with the scale based on decibels rather than normalized values as you have it set)...  -1 dB peak plus 2 dB amplification = clipping.  Something already at 0 db plus any amount of amplification will be clipped (or clipped more). 

A hard limit process is often used to apply an overall gain to as much as possible that is reduced on louder parts in an effort not to clip those.  The better programs will test to determine how much gain can be applied without clipping (you can use a trial and error test process with those to see how much can be applied before clipping the strongest parts). 

In your case I'd go back to the raw file and try to find something to declip the oversaturated portions first.  Izotope RX is among the best for that.  Deamplifying those parts will only reduce their overall output but not address any inherent clipping (i.e., if the peaks are flat-topped they will stay that way no matter how much you deamplify it). 

Assuming something works to improve those you can then see if any gain is needed for the overall file.  What I see in that shot suggests you had it all pretty well maxed out (or overrecorded) and there's no room to apply any gain other than perhaps bringing it all up that marginal bit between where the current limit is and whatever you choose as the final limit.  That would not be a hard limit but just a mathematical determination of how much could be applied to the highest peak without clipping it.  It looks pretty compressed to me...  A hard limit would just further compress it.  Music needs some air in it to breathe.  Pushing it all or all the sections to the limit just makes it more brickwalled. 

[Life In Rewind]

You might try to make more compressed/overloaded sections lower (de-Amplify) - they look like they have more RMS power - and perhaps sound louder.

The rest looks ok...

[buckster]

I've never used audacity so I don't know what if any VST plugins are compatible, but I'll point you to this free plugin that purportedly will recover heavily clipped files.  I've never had to use it myself, and I'm sure it's no comparison to Izotope Rx, but here's the link if you're interested. 

http://www.terrywest.nl/utils.html

[Marshall7]

I've never used audacity so I don't know what if any VST plugins are compatible, but I'll point you to this free plugin that purportedly will recover heavily clipped files.  I've never had to use it myself, and I'm sure it's no comparison to Izotope Rx, but here's the link if you're interested. 

http://www.terrywest.nl/utils.html

Hey, thanks for the link.  I'm going to try it out more, but at first glance it looks pretty good.

[nicolaidenmark]

Will look into it. Cheers.

[buckster]

I've never used audacity so I don't know what if any VST plugins are compatible, but I'll point you to this free plugin that purportedly will recover heavily clipped files.  I've never had to use it myself, and I'm sure it's no comparison to Izotope Rx, but here's the link if you're interested. 

http://www.terrywest.nl/utils.html

Hey, thanks for the link.  I'm going to try it out more, but at first glance it looks pretty good.

You're welcome!  It's in my "toolbox" but I've never had to use it, so it would be nice to know of any real world experience with it. 

[bombdiggity]

I've never used audacity so I don't know what if any VST plugins are compatible, but I'll point you to this free plugin that purportedly will recover heavily clipped files.  I've never had to use it myself, and I'm sure it's no comparison to Izotope Rx, but here's the link if you're interested. 

http://www.terrywest.nl/utils.html

Thanks.

That looks like a really good plugin.  I like Izo for the rare transient but have not been as enthusiastic about it in an automated mode for something that has really major continuous sorts of regions.  Might be interesting to try though can't think of anything I'd need it for right now... 

[nicolaidenmark]

The wierd thing is, even with shows that didn't clip, I've never been able to amplify with allowing audacity to clip.

[bombdiggity]

The wierd thing is, even with shows that didn't clip, I've never been able to amplify with allowing audacity to clip.

Well you should be able to, but obviously would need to pick an amplification level that is lower than the "headroom" remaining in the source. 

You'd probably be helped a lot if you convent the vertical scale on your waveform view to decibels.  Then you can see exactly how much room there is between the existing peak and the maximum. 

I assume there is a hard limit function which should do the math for you and allow you to compress any spikes a little without clipping them if you care to. 

Audacity is a somewhat crude tool and IMO not really the best for a lot of things but I'm sure one can amplify without clipping if one selects the right amplification level relative to the material.

[opsopcopolis]

The amplify plugin in audacity defaults to the amount of amplification that would bring the highest peak to 0 (which is essentially normalizing).  In general they way to do what you're asking is to lower the overall dynamic range with compression, and then compensate by upping the gain (you're loads get softer allowing you bump up the entire thing).  It definitely looks pretty heavily compressed already though, so I'm not sure that's really worth it.  Only thing I can really think of is to independently amplify the quieter sections. Definitely not the smoothest option but may be the best in this situation

[nicolaidenmark]

I have tried to highlight and amplify the quieter sections individually too but again audacity won't do it.