Remove "gong" from audio?

[blackflames]

Hello,

I have a recorded video of a speech in an auditorium. The audio is overall good, but there is this "gong" in the vocals. When the speaker speaks, there is this heavy effect in the voice. I hope you get the point when I refer to recording in an auditorium.

Audio: MP3
Software I have: Adobe Audition, Goldwave, Audacity.

What filter/effect should I use to fix the vocals? Is there any plug-in for either of the softwares that can help me equilize the levels? I am very good with using effects, so please explain in a bit detailed manner.

Thanks!

[redwudz]

You might post a short MP3 file to show us what you mean by 'Gong'. Then someone here might be able to give some suggestions.

[blackflames]

Here you go . It's a sample from a speech on pilot training.

sample.mp3

[Chris K]

The most annoying part of the distortion is around 100 Hz.

You can use a small band equalizer in the best of the audio editors you have
and move the slider at 100 (110?) Hz down to its minimum.

EDIT:

You can also use the SuperEQ command in a Avisynth script.

DirectShowSource("path_to_audio_clip", video=False)
SuperEQ("path_to eq_preset/speech.feq")

It uses a 18 band equalizer that can be set to +/- 20dB.

Here's a suggested preset... speech_preset.zip

[blackflames]

I am using Adobe Audtions. I used the Quick Filter and brought it flat to 53. That did help, but not completely remove it. Any other suggestions?

[Chris K]

I am using Adobe Audtions. I used the Quick Filter and brought it flat to 53. That did help, but not completely remove it. Any other suggestions?

Some parts of the distortion are in the spectrum of the speech and you would affect the speech to much if you try to filter that out.

[blackflames]

I am a bit confused now. Here's what the people at Audacity forum say:

http://forum.audacityteam.org/viewtopic.php?f=28&t=14107

[Chris K]

@audacity forum

The weird noises in this recording are compression artifacts.
You could reduce these in future by recording lectures at a sample rate of 22050Hz instead of 48000Hz.
(The bit rate for your WAV to mp3 conversion may be too low, this can also cause a lot of compression artifacts to occur).

Weird explanation !?.

IMO the noise is already on the recording. You should be able to check that on the original source.

[jagabo]

The audio sample you provided was 48 KHz samples MP3 compressed at 192 kbps. But at some point it was recorded at a much lower bitrate (and probably a lower sampling rate too). So in addition to reverberation from the hall in which it was recorded you have terrible compression artifacts. There is no way to fix it unless you have a higher quality source.

[Xpenguin17]

Here you go.

sample.wav.rar

[Chris K]

So in addition to reverberation from the hall in which it was recorded you have terrible compression artifacts.

The low rumble distortion is constant. I would expect that artifacts don't show up when there is silence

[jagabo]

If all you want to do is get rid of the rumble a high pass or notch filter will work -- like Xpenguin17 sample. But that's the least of the problems. What's left still sounds like shit. And I don't think it address the OP's complaint: "When the speaker speaks, there is this heavy effect in the voice".

[Xpenguin17]

I used a fine-tuned denoiser and checked the inverted output. No legitimate audio was audibly shelved. But since the rumble already masks the lower bands of his voice anyway, it sometimes does sound shitty without the noise. That's why you need the original WAV. MP3 shit-cans inaudible sound masked by louder sound, so any parts of his voice "below" the layer of noise is gone.

[blackflames]

So in addition to reverberation from the hall in which it was recorded you have terrible compression artifacts.

The low rumble distortion is constant. I would expect that artifacts don't show up when there is silence

You're right. It's only in the voice of the speaker due to the echo in the auditorium. Voices of people sitting right by me are clear.

[jagabo]

Voices of people sitting right by me are clear.

That is always the case: The difference in path length between the person sitting next to you and their voice reflected off the walls (reverb) is likely 10 to 100 fold. So the reflected signal is very weak compared to the direct signal. On the other hand the direct signal from the speaker vs the speaker's signals reflected off the walls is more like 1.5 to 2 fold. So the reflected signal is only slightly attenuated compared to the direct signal -- hence lots of reverb.

[blackflames]

So I apply a Convolution Reverb to it or something is suggested?

[lordsmurf]

It's an issue of a poor microphone, high compression recording, bad room to record in, plus all the ambient noise. Your source recording is really shrill, compression artifacts made it worse. Attached is probably as good as it will get.

sample12.wav

In Sound Forge, I removed some hum via custom paragraphic EQ settings.
In DiamondCut, I re-balanced the audio to push the "good audio" higher than the noise, slightly suppressing noise.
In Goldwave, I removed what was left of the low rumble.

[poisondeathray]

I hate to be the narc here, but this is a joke.

It's from a movie currently in the theatres. Pandorum.

Wait for the blu-ray/dvd and I'm sure it will sound a lot better :P

[aedipuss]

wow some cammer got the best around to help him. guess he hasn't heard of using the direct line audio output for the hearing impaired.