anyone here great at removing echo?

[Brett Culp]

I have an audio clip of people giving speechs at an event. It's about 20 minutes of audio.

The audio technician did NOT have an audio board with enough outputs, and my only option was to record from a shotgun mic about 10 ft away from a speaker... the layout of the room kept me from getting the mic any closer to the speaker.

The location was a very large, echoey room. The audio is clear, but it has a distracting echo.

Is anyone here skilled at improving audio with this issue? It doesn't have to be perfect, just better.

If you are skilled, please send me an email with your rate for this type of work. My email is brett@brettculp.com.

(BTW - I am not seeking advice on how to fix it myself or record differently next time. I am looking for a skilled solution.)

Thanks! :-)

[Paul R Johnson]

Could you post a short link showing the problem. The nature of the problem is about the type of unwanted sound. Some can be treated, while others can't. Having a listen would enable people to determine if they can help, or not.

[Brett Culp]

Hi Paul,

Thanks for adding this info. Here's a link to listen:

www.brettculp.com/test.mp3

[Chad Johnson]

I don't really think this is something that is fixable.

I do audio restoration, and the process of lowering the reverb would also degrade the main voice. Honestly, it doesn't sound that bad. The words are very clear, and if it's a big room it won't sound odd when played with the visuals showing such a room. It's distracting to you because you wanted to get the mic closer, but to the casual listener I think it sounds fine.

[Jim Andrada]

Generally speaking removing echo is not too successful. In this case though the echo is so consistent and continuous that it might be treatable to some extent with noise reduction. I fooled with it for a bit and echo is reduced, but there is some impact to voice quality.

I think if I were to play with it for a couple of hours I could strike a better balance between reducing the echo and degrading the voice quality. But I don't think there is any magic that will significantly reduce the echo without at least some degradation of the voice.

Original

http://files.me.com/jimandrada/cej4rq.wav

Reduced echo

http://files.me.com/jimandrada/lur3bv.wav

[Ben Denham]

From "The Ten Commandments of Sound for Picture" stickied at the top of this forum.

"10) Thou shalt accept the fact that, once and for all, there is no such thing as a "remove echo" filter."


Although the writer of those commandments does offer some hope for those dealing with echo problems-

"Guess what, Chachi, there is a solution: it's called the "shoot-it-again" filter."

I've found myself using the "shoot-it-again" filter quite a bit and I'm as guilty as the next guy of not following all these commandments but it is definitely a list that is worth aspiring too.

[Bob Grant]

To some extent echo can be removed with a downward expander. Combine that with some careful Eq and you can get a usable improvement or end up making the sound worse.

You can build a downward expander as follows. Duplicate the track and invert. You should hear nothing. Now add a compressor to one of the tracks. Careful adjustment of the compressors parameters can yield a result where the trailing echo is attentuated as it's below the knee of the compressor and cancels out.

In this case my feeling is the sound probably matches the vision and I'd leave well enough alone.

[Paul R Johnson]

We really should be clear here - this clip has no echo whatsoever! It has excessive reverberation. There seem to be typical large church/cathedral artefacts - no early reflections then really complex hard surface reflections - in effect, a wideband (well, at least in the vocal range) mess. As mentioned, expansion of the dynamic range lifts the fairly strong voice, and then the reverberation appears a little less in overall level. Intelligibility is pretty good, considering.

If it was echo, then other possibilities appear - such as perhaps trying cancellation on the distinct echos evident at the end of each phrase - but in this example, none of these echo 'tricks' will work, because there isn't any!

I'll try a few things as soon as I get a moment.

[Brett Culp]

Thanks for all the insights you have shared. I look forward to re-visiting this early next week.

Warmly,
Brett

[Jim Andrada]

Paul - you make a good point. Just how does echo differ from reverb?

[Jim Andrada]

Paul - you make a good point.

Just how would you define the difference between echo and reverb?

I always thought that from a purely physical point of view the same mechanisms were at work, but with different physiological effect in that if the reflected sound followed the original sound by some small amount it was perceived as a continuation of the original sound (reverb), but if the delay were longer it was perceived as a new sound (echo.)

It would follow (I think) that if the speaker were 50 feet from a reflective surface, then the speaker would hear an echo as would a microphone close to the speaker, but if the microphone were close to the reflecting surface, the time between arrival of the direct sound and the reflected sound would be short enough that it would appear as a reverberation. And then the speaker could be close to a reflecting surface so it would appear as a reverberation to the close mic as well as the distant mic.

Of course there could be lots of reflecting surfaces at varying distances, in which case it would appear as an unholy mess regardless of what one called it.

[Steve House]

The physical nature of reflection would be the same when creating both reverb and echos but the arrival time of the direct and indirect waves makes the difference. Reverb occurs when the reflected sound arrives at the ear or microphone DURING the direct sound. This results in the incoming delayed and phase-shifted reflected wave actually mixing with the direct sound wave in such as way as produce a modified waveform. Echo, OTOH, occurs when the time interval is long enough that one hears two distinct sounds and there's no physical interaction between the direct and reflected waves. If we were to look at the range of effects as arrival time difference increases starting from exactly coincident and in-phase, we would see it go Reinforcing-->Comb Filtering--> Reverb-->Echo. The nature of the sound also enters into it. A single impulse would always lead to an echo instead of reverb because the reflected wave would always arrive after the direct wave had ceased. OTOH a sustained violin or trumpet note could exhibit some reverb effects even if the reflecting surfaces were fairly far away since the indirect waves would be arriving as the direct sound continued.

This is why it is so hard to clean up reverb. It's not just that there's an extra sound there to be filtered or erased. The waveform of the desired direct sound has actually been modified by virtue of being mixed with the indirect sound as it was being recorded and we really don't even HAVE a recording of the desired sound to work with.