Research On Removing Room Echo/Reverberation In Progress

It's Not Hard to Get Echo
This just some info from a few articles online that have ideas that might somehow be adapted to echo/reverb removal in Audacity. None of this is mine, it's just being stored here while I play with it.That's what she said. These are just rough clips taken out of context to work from.
I haven't had any luck with this but I hope you figure it out. maybe it's just like everyone says on the forums, echo and room ambiance can't be fixed post recording. Here's what I gathered.
Reducing Recorded Echo
This question Reducing Recorded Echo This question comes up a lot. And of course, the answer is that echo must be prevented, not treated. Here's a technique (courtesy of Bob Grant) that may help. Good luck with it. Duplicate the audio track and invert. You should now here nothing. Apply serious compression to the duplicated track. Adjust threshold, attack and release times and the level of the track. The aim is so that the compressor is leaving the lower level echo untouched and compressing the wanted part. The net result (hopefully) will be that the lower level echo cancels out, leaving the wanted portion intack. You can also knock some of the bottom end out. Apply a low shelf on the buss master. Once you have everything as good as you can render it all out to a new audio track, delete the originals and reset all the FXs, just makes it easier to edit in my opinion. Do not expect a perfect result but if it's just speech you can get a big improvement. However, there actually is a method for reducing recorded echo that was described in this forum by Bob Grant. I've tried it extensively and found that how well it works for me depends how much higher the amplitude of the main signal is than that of the echo. A small amount of echo can be almost eliminated. But I was unable to improve a large amount of echo at all without seriously degrading the main signal. Also, low level intelligence in the signal tends to get lost-- part of the baby may get thrown out with the bath water. For example, consonant sounds at the beginning or end of some words from some speech may suffer. The more dynamics the main signal has, the harder it is to find the right settings. For instance, the nuances of dramatic dialog will probably not fare as well as an address by a public speaker. In the interest of having a self-contained thread on this topic, I'll quote Bob here rather than link you to his post. " ... Basically you make a copy of the track and invert it. Now you have 100% cancellation. Now apply compression to one of those tracks. Trick here is to get the knee of the compressor so that the compressor is just compressing the wanted part of the signal and not touching the lower volume echo / reverb. So now the echo will still cancell out but the compression will alter the wanted part of the track so it doesn't get cancelled. Adjust attack and decay times of the compressor to taste as well as the level of one of the tracks. Adding Eq after the two tracks are mixed as well will help. You'll never get it 100% as good as it would have been minus the echo but you can make a very signifcant improvement." I start (in Vegas) with 10 to 1 compression, 250 milliseconds of attack and 250 milliseconds of release and adjust the threshold ("knee") until the sound is optimal. Then I adjust the release downward until distortion outweighs the echo reduction benefit. I find that the process is much less sensitive to release time than to attack time. There's nothing magic about 10 to 1 compression--a wide range of ratios works. There is probably some correlation of optimal ratio to the type of signal and type and amount of echo, but I haven't tried up a lot. And of course, the answer is that echo must be prevented, not treated.
Here's a technique (courtesy of Bob Grant) that may help. Good luck with it.


Duplicate the audio track and invert. You should now here nothing.
Apply serious compression to the duplicated track. Adjust threshold, attack and release times and the level of the track. The aim is so that the compressor is leaving the lower level echo untouched and compressing the wanted part. The net result (hopefully) will be that the lower level echo cancels out, leaving the wanted portion intack.

You can also knock some of the bottom end out. Apply a low shelf on the buss master.

Once you have everything as good as you can render it all out to a new audio track, delete the originals and reset all the FXs, just makes it easier to edit in my opinion.

Do not expect a perfect result but if it's just speech you can get a big improvement. 
However, there actually is a method for reducing recorded echo that was described in this forum by Bob Grant. I've tried it extensively and found that how well it works for me depends how much higher the amplitude of the main signal is than that of the echo. A small amount of echo can be almost eliminated. But I was unable to improve a large amount of echo at all without seriously degrading the main signal.

Also, low level intelligence in the signal tends to get lost-- part of the baby may get thrown out with the bath water. For example, consonant sounds at the beginning or end of some words from some speech may suffer. The more dynamics the main signal has, the harder it is to find the right settings. For instance, the nuances of dramatic dialog will probably not fare as well as an address by a public speaker.

In the interest of having a self-contained thread on this topic, I'll quote Bob here rather than link you to his post.

" ... Basically you make a copy of the track and invert it. Now you have 100% cancellation. Now apply compression to one of those tracks. Trick here is to get the knee of the compressor so that the compressor is just compressing the wanted part of the signal and not touching the lower volume echo / reverb. So now the echo will still cancell out but the compression will alter the wanted part of the track so it doesn't get cancelled. Adjust attack and decay times of the compressor to taste as well as the level of one of the tracks. Adding Eq after the two tracks are mixed as well will help.

You'll never get it 100% as good as it would have been minus the echo but you can make a very signifcant improvement."

I start (in Vegas) with 10 to 1 compression, 250 milliseconds of attack and 250 milliseconds of release and adjust the threshold ("knee") until the sound is optimal. Then I adjust the release downward until distortion outweighs the echo reduction benefit. I find that the process is much less sensitive to release time than to attack time. There's nothing magic about 10 to 1 compression--a wide range of ratios works. There is probably some correlation of optimal ratio to the type of signal and type and amount of echo, but I haven't triedReducing Recorded Echo
This question comes up a lot. And of course, the answer is that echo must be prevented, not treated.Reducing Recorded Echo
This question comes up a lot. And of course, the answer is that echo must be prevented, not treated.
Here's a technique (courtesy of Bob Grant) that may help. Good luck with it.


Duplicate the audio track and invert. You should now here nothing.
Apply serious compression to the duplicated track. Adjust threshold, attack and release times and the level of the track. The aim is so that the compressor is leaving the lower level echo untouched and compressing the wanted part. The net result (hopefully) will be that the lower level echo cancels out, leaving the wanted portion intack.

You can also knock some of the bottom end out. Apply a low shelf on the buss master.

Once you have everything as good as you can render it all out to a new audio track, delete the originals and reset all the FXs, just makes it easier to edit in my opinion.

Do not expect a perfect result but if it's just speech you can get a big improvement.

However, there actually is a method for reducing recorded echo that was described in this forum by Bob Grant. I've tried it extensively and found that how well it works for me depends how much higher the amplitude of the main signal is than that of the echo. A small amount of echo can be almost eliminated. But I was unable to improve a large amount of echo at all without seriously degrading the main signal.

Also, low level intelligence in the signal tends to get lost-- part of the baby may get thrown out with the bath water. For example, consonant sounds at the beginning or end of some words from some speech may suffer. The more dynamics the main signal has, the harder it is to find the right settings. For instance, the nuances of dramatic dialog will probably not fare as well as an address by a public speaker.

In the interest of having a self-contained thread on this topic, I'll quote Bob here rather than link you to his post.

" ... Basically you make a copy of the track and invert it. Now you have 100% cancellation. Now apply compression to one of those tracks. Trick here is to get the knee of the compressor so that the compressor is just compressing the wanted part of the signal and not touching the lower volume echo / reverb. So now the echo will still cancell out but the compression will alter the wanted part of the track so it doesn't get cancelled. Adjust attack and decay times of the compressor to taste as well as the level of one of the tracks. Adding Eq after the two tracks are mixed as well will help.

You'll never get it 100% as good as it would have been minus the echo but you can make a very signifcant improvement."

I start (in Vegas) with 10 to 1 compression, 250 milliseconds of attack and 250 milliseconds of release and adjust the threshold ("knee") until the sound is optimal. Then I adjust the release downward until distortion outweighs the echo reduction benefit. I find that the process is much less sensitive to release time than to attack time. There's nothing magic about 10 to 1 compression--a wide range of ratios works. There is probably some correlation of optimal ratio to the type of signal and type and amount of echo, but I haven't tried
Here's a technique (courtesy of Bob Grant) that may help. Good luck with it.


Duplicate the audio track and invert. You should now here nothing.
Apply serious compression to the duplicated track. Adjust threshold, attack and release times and the level of the track. The aim is so that the compressor is leaving the lower level echo untouched and compressing the wanted part. The net result (hopefully) will be that the lower level echo cancels out, leaving the wanted portion intack.

You can also knock some of the bottom end out. Apply a low shelf on the buss master.

Once you have everything as good as you can render it all out to a new audio track, delete the originals and reset all the FXs, just makes it easier to edit in my opinion.

Do not expect a perfect result but if it's just speech you can get a big improvement.

However, there actually is a method for reducing recorded echo that was described in this forum by Bob Grant. I've tried it extensively and found that how well it works for me depends how much higher the amplitude of the main signal is than that of the echo. A small amount of echo can be almost eliminated. But I was unable to improve a large amount of echo at all without seriously degrading the main signal.

Also, low level intelligence in the signal tends to get lost-- part of the baby may get thrown out with the bath water. For example, consonant sounds at the beginning or end of some words from some speech may suffer. The more dynamics the main signal has, the harder it is to find the right settings. For instance, the nuances of dramatic dialog will probably not fare as well as an address by a public speaker.

In the interest of having a self-contained thread on this topic, I'll quote Bob here rather than link you to his post.

" ... Basically you make a copy of the track and invert it. Now you have 100% cancellation. Now apply compression to one of those tracks. Trick here is to get the knee of the compressor so that the compressor is just compressing the wanted part of the signal and not touching the lower volume echo / reverb. So now the echo will still cancell out but the compression will alter the wanted part of the track so it doesn't get cancelled. Adjust attack and decay times of the compressor to taste as well as the level of one of the tracks. Adding Eq after the two tracks are mixed as well will help.

You'll never get it 100% as good as it would have been minus the echo but you can make a very signifcant improvement."

I start (in Vegas) with 10 to 1 compression, 250 milliseconds of attack and 250 milliseconds of release and adjust the threshold ("knee") until the sound is optimal. Then I adjust the release downward until distortion outweighs the echo reduction benefit. I find that the process is much less sensitive to release time than to attack time. There's nothing magic about 10 to 1 compression--a wide range of ratios works. There is probably some correlation of optimal ratio to the type of signal and type and amount of echo, but I haven't tried
Zynaptic Unveil is very expensive but can remove echo in some cases.(like $300)

1/ Use a noise gate to chop the tail of the reverb
2/ use some good nr plugin
3/ add some lower range eq and experiment taking off the annoying highs in the fr
4/ bring it up to around -2 db
5/ experiment with a little reverb to get rid of any staccato type effect - yes you take reverb off to put on controlled re-verb
6/ sometimes you have to try this process twice..just depends.
-----------
PUT A NEGATVE 1 IN GVERB- from a comment- doesn't seem that GVERB accepts negative inputs, at least in the latest version. Not sure where this suggestion came from
---------
Noise gate plug-in for Audacity

Download and install the noise gate plug-in provided through the Audacity website, if you do not see "Noise Gate" listed under "Effect" on the toolbar. You
may need to restart Audacity after installing the plug-in before it appears on the menu.
   
Open the audio file with the echo you wish to reduce in Audacity. Click "Effect" on the toolbar and select "Noise Gate" from the list. The noise gate window is an offline effect, meaning that it will process your audio before you play it back to hear the effect.

 Set the controls of noise gate to remove echo and other unwanted noise content. Start with "Level reduction" at -100, "Gate threshold" at 30 and
"Attack/Decay" at 75. Level reduction tells the gate how much to reduce unwanted audio. The gate threshold sets the volume level at which the gate starts to
reduce sounds and the attack and decay setting affects how quickly the gate process starts and stops. Click "OK" to start the process.
   
Play the result of the noise gate process. Evaluate the effect of those settings. If there is no change to the echo, increase the threshold setting until the
echo occurring after important audio is sufficiently reduced. Reduce the threshold setting if the noise gate cuts off important audio. This process may take
several attempts. Click "Edit" from the toolbar, then "Undo" to restore your audio to its original state between attempts.
   
Adjust the level reduction and attack/decay settings to make the noise gate effect more natural, after you find an effective threshold level. Increasing level
reduction adds some echo, but you can control what level. Increasing the attack and decay time smooths out how the gate effect begins and ends. Slower
settings make the effect less noticeable.

Tips & Warnings
    You may be able to use an equalizer to "tune" a resonant echo out of a recording, but this is not a technique that applies to all cases.
    Work on copies of your original file to prevent loss of data if you accidentally over-process your audio.
-----------------------
what worked for me was to use the audition preset "sing along (drop vocals 6db)" and tweak just a bit. it did remove a lot of the obvious reverb and add a bit
of tin-y soiund to the high end.  EQ'd this and it worked ok. i'm giving the client  a clean version and one with canned background music so they can choose.
-------------------
I havefound a different way of significantly reducing room generated echo. I've done it reproducably. It leaves some but substantually helps. The process is to
apply an attenuating narrow band notch filter to the track with an attenuation of about 15db's. The exact frequency is a function of t he room accustics and
the recorded voice. The room I recored in was about ~30' x 30'. For male voice I found by experimentation the notch was at about 170hz, for the female I
found it to be about 220hz. My theory as to why this works is that most returning echos are significantly attenuated and are not picked up by the mic.
However every room/object has their own unique resonant frequency. Sound generated at that frequency is amplified and causes the preponderance of the
perceived echo. This is what is removed by the notch filter. I use Sony Sound Forge Pro which allows me to, in real time adjust both Q of the filter and
frequency. I'd like to have feedback (no pun intended)on this technique.
Love the notch filter technique. I had a video of a training in an auditorium (125 stadium seating, female presenter). I set the primary notch at 181 Hz w/
nearly 73 dB reduction. I also had notches on either side at 59 Hz and 1722 Hz and -34 dB and -34 dB, respectively. The result was great. It almost
eliminated the ring and hum. The cost was that the dynamic range on her voice took a significant hit.
-----------------
Clearly unusable unless I could chromakey myself into a big church or other large room. With my deadline approaching, I quickly started Google-ing, finding
many items that said removing echo was impossible. On another site, I saw a suggestion to use Adobe Audition's Center Channel Extractor (see the bottom
post here).
The filter is simple enough to apply; in the main menu choose Effects > Filters > Center Channel Extractor. Here's the filter itself, which I just used in its
default configuration.http://www.streaminglearningcenter.com/images/center%20channel%20extractor.jpg

........................
 https://www.google.com/?gws_rd=ssl#safe=off&q=phase+polarity+invert+for+audacity
The Phase/Polarity invert button Polarity/Phase Invert Button in Cubase
 With these, we can tweak the Attack portion of a sound and boost it, and
 then separately attenuate the Release portion of the sound – the ambience. This is a bit of brute force, but it can work. But it’s not really De-Verb…

I mentioned phase/polarity invert a few lines back…
Why would we want to do this?
What we’re going to do is flip the phase of a second, duplicate channel, then change the audio on that channel, but only where we want sound to leak through (in this case, the where the audio is that we don’t want (the reverberance of the room) , we’re going to leave that alone – the two channels will be identical at those points, will cancel out and be inaudible, The perfect tool for this?
A Compressor.
Seriously, this is all we need. Compressors modify the volume of output audio based on loudness of the input, so when the sound is loud, we’ll compress the polarity/phase inverted channel (very heavily), making the two tracks different from each other which will allow the audio to leak through from the primary track. Between the hits, where the sound is quiet, the compressor will no longer be working, so the two channels of audio will be identical and will cancel out – no audio will be heard.

The Tools
For this, I’m using a ‘free’ compressor that will run in most major audio hosts  – Stillwell Audio’s The Rocket. In reality, The Rocket isn’t free, but it does have an eternal demo mode so you can try it out. Why The Rocket? It has exactly the features we want – superfast Attack, reasonably variable release (enough for us, most likely) and huge compression ratios.

How to remove Reverb – The Method


Step 1. Duplicate the track. Here you can see we ended up with AKGL & AKGR, and another pair of tracks, Copy of AKGL and Copy of AKGR. These tracks are identical.

Step 2. Send your main pair of Overheads (AKGL and AKGR) to a Group Channel. Here you can apply any EQ or Compression that you want on the original sound. When you’re happy with that, we can get on with getting rid of that Ambience.

Step 3. Duplicate that Group Channel, give is a sensible name, and the route your COPY tracks (Copy of AKGL and Copy of AKGR in this case) to this new channel. This channel should sound Identical to your previous Group – Excellent!

Step 4. Click the Polarity/Phase button on the new Channel. The sound from your Overheads should vanish. This is a good thing.

Polarity/Phase button on second  channel
Step 5. This is the last, but most importnat step – it’s going to require a bit of tweaking, but I’ll give you a starting point. any time I mention mucking around with the channel volume – muck around with the ‘Parallel Compression’ knob in The Rocket

Stillwell - The Rocket - De-Verbing like a Pro
Turn the Attack to its lowest setting (to the Left). Leave the Release quite high to start with (we’ll be tweaking that anyway – 500ms will do), whack the Ratio on 20:1, and then select the GR (Gain Reduction) Meter on the right. Now, drag that threshold DOWN. Really… DOWN.
We want to be hammering the audio. Don’t touch Detector HPF or Parallel Compression (previous note taken into account .
 What you should notice, if you play the audio now, is that you hear something. In fact, you’ll probably hear a lot. What’s happening is that, although the two channels are of inverted polarity/phase, the compressor is vastly changing the inverted channel making it different (and much quieter) than the original channel. They will no longer cancel out. We now need to tweak to get rid of the reverb.

The Gain Reduction meter should be flying to the left, and probably hanging around there. We don’t want that – we need to shorten the release time so the compressor recovers faster. We want the compressor to hit immediately (hence the short attack) but stop compressing during the ambience. For an extreme effect, turn the release down to about 50ms, then edge it back up until you get the kind of ambience reduction you’re after.
I
 went for fairly (2) subtle reduction on these tracks but you can take it a lot further. Here, try playing the two players in parallel (press play on the first, then play a second or so later on the second) and you can do a comparison. Back in your DAW, for a bit more ‘tweakability’, you could go more extreme on the overall settings and then blend it in more subtly by pulling down the volume of the inverted track in the Mixer.

This also opens up a lot of different avenues.. Here, we’re working with a classic style compressor – we have static attack and release times, and a variety of ratios to choose from.

But what if we chose a limiter with program dependent release? What about the compressors with program dependent or interesting non-linear release curves?

I’ve yet to try any commercial De-verb products, but I imagine they also struggle with long, relatively quiet sustained sounds with minimal dynamics. I can’t immediately think of any simple, non-spectral processing dealing with that
.
So, there you go – removing reverb for free. Any thoughts?
Footnotes:

(3) Plenty of options here.

turn off any saturation provided by the plugin first – it’ll muck around with your phase cancellation. Finally, there’s no harm in using a DAW bundled compressor…
    It’s polarity that’s being inverted, not phase.

    why is this any different to using a noise gate? Or have you basically devised a way to implement a compressor as a noise gate here? It doesn’t clean up quite as well with the detailed waveform of speech i guess, because an extended loud section has the “early reverb” overlaying the later loud part of the loud section
  
        You’re absolutely right, it’s not a panacea – it’s another tool in the arsenal (and essentially, yes, we’re using compression and polarity inversion as a noise gate).  Speech (early reflections are always going to be a problem