How to adjust uneven sound on video?

cdkeeka wrote on 10/14/2015, 9:44 AM

A friend has given me his ten year old wedding VHS video (taken by another friend) to transfer to DVD. Unfortunateley though the picture is quite normal for VHS, the sound is uneven. Every few seconds, the sound becomes louder for a few seconds then back to normal again. As far as he can remember, the sound has always been like this. Either the original tape was faulty, or sound was damaged in storage. "Normalise" setting either in MEP sound editing or Sound Cleaning Lab does not make any difference ( I have VSCL version 1.0 as the latest version, quite frankly as far as I can see on the forum, the latest version is not an improvement). There are also various clicking sounds, either due to fiddling with the camera or the lense cap hitting the camera body, and also what sounds like a airconditioner hum. These last two I have managed to reduce or remove using VSCL.

My question is how to fix or equalise the uneven sound; to correct the sudden burst of increased volume?

Is there a plugin I can use, or do I mannualy have to adjust the volume all the way through?

Thanks for any advice anyone can give me.

Comments

browj2 wrote on 10/16/2015, 11:10 AM

Hi,

VSCL 1 has spectral cleaning. Use it to remove or reduce sudden unwanted sounds. It takes some practice.

Air conditioner hum - you would need to try to find the offending frequency and try reducing it using the equalizer. You can also try the FFT filter and or the denoiser. You have some reading ahead of you and some trials. You want to try to keep the good sound and filter out the bad.

For the volume, in VSCL, make sure that you have normalized the recording, then under Enhance, DeClipper - remove distortions - fine-tune this and the following, and Dynamics - cutoff noise or strong noise - this is a compressor and you should do some compression. More reading.

Then you can either adjust the volume manually by adjusting the volume curve or semi-manually using volume automation in MEP. More reading. For this, you open the mixer, click on auto for the track that you want to adjust, start playback and adjust the volume sliders up and down. This adjusts the volume curve on the fly, takes practice and you will have to make fine-tuning adjustments. You should watch the waveform in the track during playback so that you know where the spikes occur and you can start reducing the volume as the playback marker gets there.

Basically, there is no set way to do this. You have to learn what the tools do and then try the right combination and settings to get the results that you want.

BTW, VSLC 1 has the detailed Spectral Cleaning that was discontinued in ACL and AML - even the best ones available now don't have this detailed and excellent tool. Make sure to keep VSLC 1 on your computer. I have AML 2014 Deluxe and it has excellent tools and the 2016 version is even better. But none have that detailed spectral cleaning tool.

HTH

cdkeeka wrote on 2/13/2016, 10:52 AM

Thanks for your answer and sorry for the delay in answering.

You were right - there was a lot of trial and error trying different settings and options. The biggest problem was the sound, but the picture also needed adjusting - contrast, colour, etc.

It took me quite a few all night sittings and weekends, as the full tape is nearly 2 hours long, which included parts of the drinks before the wedding, the full wedding, then the lunch, speeches and the dancing !

I found splitting the tape into logical chapters and working on each part seperately was easier - also each one needed slightly different processing. It also meant that they could be tested seperately by burning onto a smaller Re-Writable disc (less time) so I could hear what the final version sounded and looked like

I was hoping that some VST plugin would help, tried a number - I found 'Blockfish' made some difference but still left some of the sudden burst of volume.

I then used the main screen of VSCL which showed me the variations of the volume and manually decreased the louder parts to be in line with the lower volume. However this caused a problem as the whole volume then became too low.

So back to the original file and manually increased the lower volume of the spoken parts up to be in line with the higher volume. This was better but the background airconditioning sound became uneven.

Trying to remove the airconditioning hum with the denoiser also removed some base from the dialog, which made it sound computerised, after a lot of trial and error, I only reduced it slightly and each variation had to be processed seperately.

Another side effect was the the clicking and whirring of the rewind motors of the analog film cameras of the time used by some of the guests; originally these were somewhat hidden behind the louder airconditioning hum - these now became louder and clearer. This is where the good old spectral cleaning came in very handy as I was able to remove some of the clicking and whirring. There where hundreds of low and loud clicks and whirrs so I only removed the ones I thought were causing a disturbance to the important parts of the dialog - even then I must have removed about 200 of them.

The adjustment of the volume and the removal of the disturbance took most of the time.

Finally, declip and normalise the sound of the different chapters so the whole disc would sound the same, added a low background crowd hum from my media pool where there was no important dialog, using the compressor to bring up the dialog where necessary, design the disc menu to show the various chapters and in the end burn the whole project onto a DL disk.

In all it took me two months of weekends and quite a few all nighters.

I finally handed the completed work to him in mid December.

I did get a "thankyou" but I wonder how much it would have cost him if he had given it to a professional - maybe there was an easier way.

Once again, thanks for your advice

Regards

browj2 wrote on 2/13/2016, 12:18 PM

Hi,

Thanks for getting back. Very interesting. Only those who actually go through this understand the work involved. The viewers have no clue and will even say that they see and hear no difference (but it's very good) - unless you put before and after side by side.

You have some other tools that you may or may not have used, including EQ and compression, but intermittent volume changes is quite challenging, as is getting rid of background noise. There is no easy solution, just a lot of effort.

Add in the video and image discrepancies and the process doubles. Getting white balance and colour evened out from one scene to another, especially with different cameras adds to the complexity, plus shaky video, some annoying person talking in the background during a song, etc. Then there's editing to make everything coherent. It's endless.

I did a similar video for my niece, but without the sound problems that you had. The complexity came when she asked for a certain person, the best man, to be basically cut out, for a special version. Hours or work on that one. My wife also sang part of a song, a capella, including the beginning and ending. In the video, there is piano and she sings the entire song, with cutaways and still images added during most of the song. Everyone now thinks that there was a piano and that she sang the entire song live. Even the lips are in sync at the beginning and end.

For VHS, I find that there is a lot of video noise, so I use Neat Video regularly to smooth it up. It makes a huge difference, but the rendering time is huge. I also use it on DV files taken in low light where there is a lot of grain in the video.

I have been working on VHS recordings of some of my wife's shows and in one there is recurrent thumping noise that is quite noticeable in the lower volume passages. Like you, I used Spectral cleaning and there were about a hundred or so thumps to be removed (still not finished) without affecting the music. Good thing this only happened on one video of about 10 minutes.

I didn't mention it, but Melda Productions has an equalizer, MDynamicEQ and MAutoDynamicEQ. I'm still in the learning process, but you can surgically remove or lower a small frequency range where there is offending noise. As far as I understand, MAutoDynamicEQ automatically reduce the sound only if there is a sudden gain. The trick is to find the offending frequency or frequencies. I still have quite a ways to go before I understand how to use this properly, and in which circumstances.

Also, Magix Audio and Music Lab (not Audio Cleaning Lab) has 5 essential FX filters that are quite useful.

You did well to break up the video into logical pieces and work on them, export them and review them. It's quite annoying to get to the end to then find out that what you did sounded and looked good in preview but was bad in the export.

I have found recently that for some reason, the cleaning tools used directly in MEP or VPX7 give me clipping in the export. This may be due to the use of the Melda EQ. I mixed down the sound, exported, cleaned up the sound using Melda and other tools in AML, exported/imported into the video, and it worked fine upon export.

So, congratulations on finishing the video and knowing when to stop. I find that a video and audio can always be improved, but knowing when enough is enough becomes challenging. Usually a deadline helps - they're coming for supper Saturday night, so the video had better be finished by then.