Tuesday 5 March 2013

REMOVING ECHO FROM AUDIO

Before you think this is going to cure all those nasty echo problems, it won't, but it may get you out of a spot when you've undertaken that crucial once in a lifetime interview in a  tiny, bare dressing room with that Hollywood star who simply had "no time darling" to allow a more suitable venue to be found - too bad, you have your interview but it sounds rough - fine for the news media and stinky old You tube, but rubbish for that documentary you have to get out by the middle of next week.

This will NOT remove the echo (liveness) from the actual words spoken by our hero, but will assist in dumping the awful ambience that makes it sound like you were in a public convenience during your chat.

First off, your interview will need to have been properly recorded, not using a handycam 20 metres away, there's no fix for poor craftsmanship - 

You will need a fairly good audio platform to do this - I used Audition recently to solidify some voice recording, and "pull it forward" - creating dialogue which, whilst not perfect, was certainly absolutely useable in the context of the documentary for which it was recorded.

Geeez, I forgot to finish this post! 

1) Open up Audition, import your audio and set to multitrack view.
2) Place your clip on track 1 and copy to track 2 as well, directly below, and set the lower track phase change on.
3) Play the session - you should have silence as the 180 degree phase difference
should entirely cancel out all the audio
4) Now add the Multiband Compressor to the top track and begin to play with
the settings - use the "ganged" setting to start with. Be very careful, take your time, this is not an instant fix.
5) The resulting audio will have all the echoing gaps between the words removed but without the pumping and clipping associated with gating, and the overall dynamic will be reduced - but you will still hear the "live" sound of the echo in the voice - this you have to put up with.

Adjusting the levels within the Compressor very slowly, and maybe setting the master gain in the Compressor to +1.00 will add a little ambience back into the mix.

 
Two tracks with phase change = no audio on playback


 6) When happy, send both channels to a Bus (mono) This will allow you to add further EQ and noise reduction to the stream without upsetting the echo removal process. 


Have the limiter off, carefully adjust levels (ganged is easiest to start with) and
play with the overall gain for ambience mix.

At this stage add EQ to the mix-down Bus - I usually use 30 band, as well as De Esher and or De Esser - if needed. 

As the upper middle, and top response get very rounded off in the processing to dampen the echo, it's a fine line between acceptable, then too much EQ or additional processing, least is always best, so go carefully, make a file and test in your project - I often make several versions to get the best results on different sections of the original recording.

adding a bus to create a downstream mono mix for additional EQ and Noise reduction
Have fun, hope this helps!