echo removal

I am working on an skype interview , but echo on it I can’t remove, Does anyone can help me, I am very appreciate this.

thank you

Adding reverb/echo is easy. Removing reverb/echo is very hard. Recording studios go to a lot of trouble and expense to ensure that recording do not have unwanted echo/reverb because avoiding the echo in the original recording is far more effective than trying to remove it later.

Which voice is it that you are interested in?
There is a big difference in the levels of each voice, and that will make processing the track much more difficult.
Is re-recording the interview an option?

Thank you for reply Steve,
I think re-recording is not an options, as the interviewer would be not available for that.
I think the man voice that sound terrible that will need to fix, any suggestion for that? any step should I do
even it is difficult, I have no option to fix this for hours

thank you

I thought that might be the case.
I can make some improvement to your(?) voice, but not much improvement to the man’s voice.

There are several different problems with the recording of the man’s voice: There is the echo, but also there is damage to the sound quality due to the data compression that is used by Skype, and the volume is a bit low. When trying to improve one of these issues it tends to make the other issues worse.

The best that I can suggest is that you use just the left channel of the recording - the man’s voice is a lot clearer on the left channel than the right channel.
To do that, click on the track name (top of the panel on the left end of the track), then from the drop down menu, select “Split Stereo to Mono”.
You will now have two mono tracks.
Delete the lower track by clicking on the [X] in the top left corner of the track.

heard better, how about my voice, how to edit that?

thank you so much

Editing your voice without making the man’s voice worse will be tricky because they are on the same track. You would need to manually mark every occurrence of your voice and then move them to a new track so that you can process your voice separately from the man’s voice. Can you be bothered to spend that much time on it?

For future recordings, use a microphone quite close to your mouth, but with a pop filter between your mouth and the microphone. Also, ensure that the place that you are doing the recording is quiet and has plenty of sound absorbing materials around (soft furnishings, carpets, coats hanging up, rugs on walls, anything like that will help - Bedrooms tend to be much better acoustically for recording vocals than kitchens or bathrooms.)

Ok, I am ready to spend hours on this. maybe I will duplicate this track and silent part when man talking…

please tell me your idea to edit my voice. also thank your for suggestion on “how to record”

thank you

Before you embark on a long job, try this on a short section of your voice and decide if the small amount of echo reduction is worth the effort:

Have you ever used the Noise Removal effect? If not, please read this first: http://manual.audacityteam.org/o/man/noise_removal.html

  1. Use the Right channel for your voice (the left channel for the man’s voice.
  2. Select a section that contains your voice and with the Noise Removal effect, capture a “noise profile”. Normally when capturing a noise profile with the Noise Removal effect you would select a section that has only noise and none of the voice that you want to keep, but in this case the “noise” that you want to remove (the echo) and the sound that you want to keep (your voice) are the same thing. The echo IS your voice, just quieter because it has bounced off a few walls before reaching the microphone.
  3. Apply the Noise Removal effect with these settings:
    Untitled.png

Here is a “before and after” example of this echo reduction:

Since you used The Voice as the Profile why does the voice itself not vanish?
Koz

Good question. It’s because the “Sensitivity” control is set very low. Thus the Noise Removal is looking to remove sounds that are “like” the voice, but quieter. The echo/reverb is very much “like” the voice (it IS the voice) but is quieter than the direct voice sound.

Some of the direct voice does get removed, so this trick needs to be used in moderation.

this trick needs to be used in moderation.

But better than we used to do where we couldn’t do it at all.
I can’t wait to try it out on my “recording in the kitchen” sound clip.
Koz

thank you for your help Steve, it works… :smiley:

Seems like this could be a useful tip for the Wiki at least.

Is the increase in frequency smoothing / reduction in attack / decay part of the general recommendation, or specific to this voice?


Gale

Smoothing may be the key to this. What that control actually does is define the amont of voice that makes it through unaffected. Koz

If “Smoothing” is too low you will hear the typical metallic artefacts of FFT processing.
I just increased the smoothing gradually until there was no appreciable improvement regarding that metallic bubbly sound.

The key to this is the “Sensitivity” control. That controls how much of the voice “slips through the gate”. It works like a “threshold” setting on a noise gate. The higher the setting, the less gets through. The idea is to set the “threshold” (Sensitivity) so that the direct voice gets through with little change, but that the reverb tail (echo) is attenuated. In effect this is working like a noise gate that is tuned to the frequency spectrum of the voice.

“Attack/decay” controls how fast the gate opens/shuts before/after the level crosses the threshold. If the input has a very long reverb and you want to make it shorter, but not to stop it altogether, you can set this to the length of reverb that you want and the reverb will decay from the level at which it crosses the threshold to “nominal silence” in the decay time set.