Hello,
I have an Echo effect on an original Audio recording in a conference room.
Please, how can I suppress/remove ECHO effect with Audacity ?
I know how to suppress background Noise with Noise Reduction but Echo no…
Thanks for your help.
Generally, you don’t, because it is hard to do.
If you want to try it in Audacity you can try centre isolation in Audacity’s Vocal Reduction and Isolation effect, or try Noise Reduction using a small area of reverb only as the Noise Profile.
Beyond that, it usually means spending money on a commercial “deverb” plugin.
If you want, you can attach a sample in WAV (raw recording before applying effects to it) so we can hear how bad it is. See How to post an audio sample.
Gale
Hello,
If you want, you can attach a sample in WAV (raw recording before applying effects to it) so we can hear how bad it is
here is my audio wav sample vocal 044 KN1
A@
Hello,
If my first audio sample is too short here is a second sample : 
https://archive.org/download/Vocal044KN2/Vocal%20044%20-%20KN2.wav
Thanks .
A@
Hello Gale Andrews
If you want, you can attach a sample in WAV (raw recording before applying effects to it) so we can hear how bad it is
have you seen my sample, please ?
can you do something ?
thanks.
Actually it’s not my job to do that. We are trying to prepare and test a new Audacity version for release, which is my job.
I’ll ask if one of the audio engineers are willing to look a it. If it is really reverb, they will probably try exactly the same as I already suggested, so of course you could do that yourself.
Gale
I don’t think that is possible. There’s just too much reverberation to be able to make any significant improvement.
For future recordings, you need to get the microphone much closer to the person speaking.
I think you’re stuck. Audacity can’t help big room echoes.
This is the reason news events and interviews always go to very serious effort to get a microphone close to the performer.
If there was a way to clean that up, they would just use the microphones on top of their cameras.
There is a way to shoot that.
But if you’re far enough back, even that’s not going to work.
The desperation method is have a performer read the text in a studio and then add effects to make it sound like a big room. That way you can control the sound.
Koz
Even if this is a video, you can have the voice performer listen to the original on headphones as they read the text in the studio. You should be able to record it several times and edit the sound so the lip sync lines up. That’s looping. Not all Hollywood sound shoots work right the first time.
Koz
OK I could not resist spending two minutes on it being told it was beyond hope. If it is mono you can only choose Noise Reduction not Vocal Reduction and Isolation.
In the short file, drag-select from about 0.67 seconds to 1.05 seconds. In Noise Reduction, take Noise Profile, Edit > Select > All, Noise Reduction at 26, 21, 11.
Boost the audio above 5000 Hz in Equalization (drag the curve upwards above 0 dB).
Effect > Low Pass Filter… at 5000 Hz, 12 dB Rolloff to reduce audio above 5000 Hz, though I could have done similar in Equalization.
I have no idea if anyone thinks that “better”. You can try yourself.
Gale
We can always create “different.” You can also make a change by using The Whole Show as profile and apply that. For some odd reason that sometimes works…poorly.
Koz
I have no idea if anyone thinks that “better”.
It’s clearer. Some of the mush is gone. Of course not being able to speak Farsi (or whatever that is) is a problem.
Did you sample a trailing echo? Is that what’s in your profile? He stopped for a breath and you snagged it?
Koz
Yes near as was available.
Gale
That’s borderline witchcraft. There’s no reason to believe the echoes will have different timbre, harmonics or overtones than the straight voice, just rearranged in relative volume. Nothing in the room is going to make new tones.
Who makes echo suppression software? It was once suggested that Convolver techniques could be used to suppress echoes, but I repeatedly tried to trace that suggestion to an authoritative source and got lots of cyclical references or pathways petering out in a cow pasture.
“No, no. I never said that. I think John got that to work once…” > cow pasture.
Koz
It’s possible the frequency changes on their own are responsible for much of the cleanup. Of course you can try it yourself.
Gale
“Absorption”, “diffusion”, “resonance”.
Absorption: Soft furnishings, human bodies, curtains, even doors and walls, tend to absorb high frequencies much better than lower frequencies. Have you ever noticed the “whump, whump, whump” coming from a club / bar / car that has electro / dance music inside? The higher frequencies have been absorbed.
Diffusion: Higher frequencies are more ‘directional’. Lower frequencies scatter and spread out more easily. One of the reasons that it’s harder to hear what a person is saying when they turn away from you is that you lose the “T’s”, “F’s” and “S’s”.
Resonance: Place two hard surfaces parallel to each other and sounds that have a wavelength that is an exact multiple of the distance between those surfaces, will bounce easily between the walls, creating what is known as “standing waves”. This is what creates the distinctive reverberation in corridors and tunnels.
Every room has a distinct and unique ‘colour’ (color), due to the combination of absorption, diffusion and resonance. One of the primary aims for acoustic treatment of a recording / performance spaces, is to make the colouration “neutral” (the acoustic equivalent of “grey”). To see the true colours in a painting / work of art, it needs to be illuminated by “neutral” light (no colour stronger than any other colour). Similarly, to hear the true frequency / harmonic content in sound / music, the acoustic properties of the room need to be neutral. Colouration of sound by a room is not necessarily a bad thing - certain performance venues are favoured for certain types of musical performance because their acoustic colour enhances the experience of that type of music, much in the same way as skin tones are enhanced by candle light.
Yes. But I said the room is not likely to generate foreign tones by itself absent something that rings such as striking a bell. The bell will always generate its favorite tones no matter who or what strikes it, but given the absence of bells in the average auditorium, that effect is not likely.
The relationship of the existing tones to each other will certainly change.
All that and it doesn’t explain the odd ability to marginally improve the performance by simply using the whole shebang as a Profile.
Koz
Walk into the show just before the performance and clap loudly once. Theory has it we should be able to subtract the room from the show. I suspect the awkwardness of doing that is the only thing preventing success.
Koz
Yes, I think John got that to work once… (> cow pasture)
That won’t work in Noise Reduction like it did in Noise Removal because Sensitivity is not a straight dB reduction now.
So in Noise Reduction, select a reverb tail between words or notes for Noise Profile.
Gale