Another attempt to fix this awful file at DVinfo.net

Kathy Smith · March 3rd, 2016, 03:25 PM

I have posted this before a while ago. I edited this just now. It's horrible in terms of the white noise. Could someone give this another crack and see if it's possible to do something with it. I'm not looking to do something that will sound like there is no problem. Just some noise removal, bringing gain up and some EQ. I have failed to do a good job as I feel that once I do EQ I'm destroying the file and there is too much sibilance.

Here is the file if someone is willing to take a look at it. I would be very grateful.

Thank you
https://www.dropbox.com/s/m6gsldv04f...o_Bad.wav?dl=0

Greg Miller · March 4th, 2016, 08:36 AM

Clearly this has been edited. There is a telltale dropout at 1:15.783, and I believe I hear a few other indications, too.

And both channels are bit-for-bit identical. Was it actually recorded this way? Or is this not the original?

Also, it may be my imagination, but I think I hear a few high frequency artifacts here and there. Was this originally recorded in some compressed format (e.g. MP3)?

Also, the voice sounds thin to me. Of course it may be this particular speaker, but it seems to be unnaturally lacking in LF to me. And I almost feel as if there's some IM distortion with some lower frequency that's not in the file. As if there was bad hum originally, but that has been removed, leaving just the IM. (But that's a long shot. It's just that I hear this gravelly sort of quality to the voice, and I'm not convinced that it's the speaker's natural voice.) Do you have a good recording of this same voice?

Just out of curiosity, I'd like to know what all went wrong here. Can you describe the history of this recording? What recorder was used? What was the original file format, bitrate, etc.?

Is there any slim chance that there is an original generation of this file that sounds at all better?

Kathy Smith · March 4th, 2016, 09:01 AM

Quote:

Originally Posted by Greg Miller

Clearly this has been edited. There is a telltale dropout at 1:15.783, and I believe I hear a few other indications, too.

And both channels are bit-for-bit identical. Was it actually recorded this way? Or is this not the original?

Also, it may be my imagination, but I think I hear a few high frequency artifacts here and there. Was this originally recorded in some compressed format (e.g. MP3)?

Also, the voice sounds thin to me. Of course it may be this particular speaker, but it seems to be unnaturally lacking in LF to me. And I almost feel as if there's some IM distortion with some lower frequency that's not in the file. As if there was bad hum originally, but that has been removed, leaving just the IM. (But that's a long shot. It's just that I hear this gravelly sort of quality to the voice, and I'm not convinced that it's the speaker's natural voice.) Do you have a good recording of this same voice?

Just out of curiosity, I'd like to know what all went wrong here. Can you describe the history of this recording? What recorder was used? What was the original file format, bitrate, etc.?

Is there any slim chance that there is an original generation of this file that sounds at all better?

Greg,

The noise (the hum you are describing) has not been removed yet. Do you hear the horrible white noise?
I do not have a recording of the same voice.
What happened was that the recording device was set to phantom power which was not necessary hence the noise.
Is there a way to salvage this? I'm not looking for a pristine clean up, something that I can simply just "get away" with, once I add some music.

Greg Miller · March 4th, 2016, 09:52 AM

I do not hear any hum, and I do not see any if I look at a spectral analysis.

Where do you hear hum?

Of course I hear the white noise. I'm surprised you didn't hear this in your headphones when recording the track.

But after the white noise is gone, the voice still sounds terrible. I hear a rather "growly" (for lack of a better term) quality to the voice. It's very noticeable in the following places (especially if you loop them):

the syllable "I" from 12.84 to 13.24,
"lea" from 17.34 to 17.66,
"oo" from 32.40 to 32.75,

I said, identifying it as IM distortion is a bit of a long shot. It's hard to tell what I'm hearing under all that white noise. But when I try to remove the noise, then the bad quality of the voice becomes more apparent. Maybe some of the "badness" is created by the heavy NR. Or maybe the phantom voltage caused the mic to be non-linear and generate the distortion there.

To clarify: are you saying that this file has not been filtered, processed, or altered in any way, except for some cuts you made to edit it?

Rick Reineke · March 4th, 2016, 11:36 AM

This is freak'n awful.. I don't hear any hum either, just loads of hiss and other artifacts. I was able to attenuate the hiss w/ multiple passes of NR, steep LP filters, expand it a few dB, add some body, de-ess, manually limit , among other surgical fixes here and there. It's still far from 'good'. Usable? That's subjective..

https://www.dropbox.com/s/tpuizuuu1w...essed.mp3?dl=0

Bernie Beaudry · March 4th, 2016, 01:06 PM

Hi Kathy,
I've been learning RX 4 so I took your file on as a challenge. I got rid of much of the white noise which uncovered some artifacts in your audio. It sounds like the back ground was getting hit with some sort of auto gain or something. I hear bits of stuff but with music I don't think you'd hear it. I did two passes of Spectral Denoiser, then eq, another pass of Spectral Denoiser, then Declip. I then eq'd one of the sections that had a duller sound than the rest.
Here you go. Tell me what you think.
https://www.dropbox.com/s/rmqeh3aiv4...20fix.wav?dl=0

Greg Miller · March 4th, 2016, 01:06 PM

It's a challenge, and that's putting it politely.

Greg Miller · March 4th, 2016, 01:11 PM

Quote:

Originally Posted by Bernie Beaudry

I hear bits of stuff

The "musical" tones and echo-like background noise you hear are artifacts created by using too much spectral noise reduction, or NR with inappropriate settings. They weren't there in the original. You can also hear a slight "robotic" quality to the voice, from the same cause.

Bernie Beaudry · March 4th, 2016, 05:13 PM

Quote:

Originally Posted by Greg Miller

The "musical" tones and echo-like background noise you hear are artifacts created by using too much spectral noise reduction, or NR with inappropriate settings. They weren't there in the original. You can also hear a slight "robotic" quality to the voice, from the same cause.

You are right on. I was making light passes but the noise is so intense and broadband it was still working too hard. I did another go.

Bernie Beaudry · March 4th, 2016, 05:19 PM

Here's another version: https://www.dropbox.com/s/xl2cym5jp2..._Bad3.wav?dl=0
The problem with this one is the voice still is getting a bit robotized.

Christopher Young · March 5th, 2016, 03:11 AM

Yes a problematic little file indeed. It has various wide band artifacts spread throughout its spectrum, I also think there is a bit of flanging going on in there some how? How come who knows. As previously mentioned when you try to clean it up it gets a bit robotic. I found the best I could get out of it was to attack it pretty hard on NR then give it a bit of normalizing compressing and limiting. Finally I added back -42dB of White Noise as it was sounding too dry. Best I could do with in the time to hand.

https://www.sendspace.com/pro/dl/nikklv

Roger Shore · March 5th, 2016, 05:00 AM

Quote:

Originally Posted by Greg Miller

It's a challenge, and that's putting it politely.

I thought I'd have a go myself, and having got pretty poor results using all the 'usual tricks', I realised just how much of a challenge this one is.

I have to say I think the sample you attached to post #7 is pretty impressive.....better than I thought could be got from this source!

Kathy Smith · March 5th, 2016, 08:07 AM

Quote:

Originally Posted by Greg Miller

I do not hear any hum, and I do not see any if I look at a spectral analysis.

Where do you hear hum?

Of course I hear the white noise. I'm surprised you didn't hear this in your headphones when recording the track.

But after the white noise is gone, the voice still sounds terrible. I hear a rather "growly" (for lack of a better term) quality to the voice. It's very noticeable in the following places (especially if you loop them):

the syllable "I" from 12.84 to 13.24,
"lea" from 17.34 to 17.66,
"oo" from 32.40 to 32.75,

I said, identifying it as IM distortion is a bit of a long shot. It's hard to tell what I'm hearing under all that white noise. But when I try to remove the noise, then the bad quality of the voice becomes more apparent. Maybe some of the "badness" is created by the heavy NR. Or maybe the phantom voltage caused the mic to be non-linear and generate the distortion there.

To clarify: are you saying that this file has not been filtered, processed, or altered in any way, except for some cuts you made to edit it?

Greg,

I did not do anything to the file other than making cuts and yes it was recorded into both channels. I was not the one who recorded this. I'm just editing it and trying to make the best out of it.
I do not hear any hum, I thought you said that the file sounds like some hum was removed. I thought that by hum you meant noise in general and said that nothing was removed.

Kathy Smith · March 5th, 2016, 08:08 AM

Quote:

Originally Posted by Roger Shore

I thought I'd have a go myself, and having got pretty poor results using all the 'usual tricks', I realised just how much of a challenge this one is.

I have to say I think the sample you attached to post #7 is pretty impressive.....better than I thought could be got from this source!

I agree Greg's file sounds best.

Greg Miller · March 5th, 2016, 09:06 AM

Sorry for the confusion about the hum (or non-hum). I was just looking for an explanation for the strange vocal distortion that I seem to hear in several spots.

How sad that there is no earlier generation or better version of this file. :-(

Out of curiosity, was it recorded with a wired mic, or with wireless? I hear a little noise burst, just after "... all the way around it" and I wondered whether that was an RF hit. Or maybe it's a clothing rub ... the white noise really obscures what's going on there.

March 3rd, 2016, 03:25 PM	#1
Kathy Smith Major Player Join Date: Oct 2010 Location: NY, NY USA Posts: 519	Another attempt to fix this awful file I have posted this before a while ago. I edited this just now. It's horrible in terms of the white noise. Could someone give this another crack and see if it's possible to do something with it. I'm not looking to do something that will sound like there is no problem. Just some noise removal, bringing gain up and some EQ. I have failed to do a good job as I feel that once I do EQ I'm destroying the file and there is too much sibilance. Here is the file if someone is willing to take a look at it. I would be very grateful. Thank you https://www.dropbox.com/s/m6gsldv04f...o_Bad.wav?dl=0

March 4th, 2016, 08:36 AM	#2
Greg Miller Trustee Join Date: Jan 2011 Location: Pennsylvania Posts: 1,786	Re: Another attempt to fix this awful file Clearly this has been edited. There is a telltale dropout at 1:15.783, and I believe I hear a few other indications, too. And both channels are bit-for-bit identical. Was it actually recorded this way? Or is this not the original? Also, it may be my imagination, but I think I hear a few high frequency artifacts here and there. Was this originally recorded in some compressed format (e.g. MP3)? Also, the voice sounds thin to me. Of course it may be this particular speaker, but it seems to be unnaturally lacking in LF to me. And I almost feel as if there's some IM distortion with some lower frequency that's not in the file. As if there was bad hum originally, but that has been removed, leaving just the IM. (But that's a long shot. It's just that I hear this gravelly sort of quality to the voice, and I'm not convinced that it's the speaker's natural voice.) Do you have a good recording of this same voice? Just out of curiosity, I'd like to know what all went wrong here. Can you describe the history of this recording? What recorder was used? What was the original file format, bitrate, etc.? Is there any slim chance that there is an original generation of this file that sounds at all better?

March 4th, 2016, 09:52 AM	#4
Greg Miller Trustee Join Date: Jan 2011 Location: Pennsylvania Posts: 1,786	Re: Another attempt to fix this awful file I do not hear any hum, and I do not see any if I look at a spectral analysis. Where do you hear hum? Of course I hear the white noise. I'm surprised you didn't hear this in your headphones when recording the track. But after the white noise is gone, the voice still sounds terrible. I hear a rather "growly" (for lack of a better term) quality to the voice. It's very noticeable in the following places (especially if you loop them): the syllable "I" from 12.84 to 13.24, "lea" from 17.34 to 17.66, "oo" from 32.40 to 32.75, I said, identifying it as IM distortion is a bit of a long shot. It's hard to tell what I'm hearing under all that white noise. But when I try to remove the noise, then the bad quality of the voice becomes more apparent. Maybe some of the "badness" is created by the heavy NR. Or maybe the phantom voltage caused the mic to be non-linear and generate the distortion there. To clarify: are you saying that this file has not been filtered, processed, or altered in any way, except for some cuts you made to edit it?

March 4th, 2016, 11:36 AM	#5
Rick Reineke Inner Circle Join Date: Aug 2008 Location: New York Posts: 2,038	Re: Another attempt to fix this awful file This is freak'n awful.. I don't hear any hum either, just loads of hiss and other artifacts. I was able to attenuate the hiss w/ multiple passes of NR, steep LP filters, expand it a few dB, add some body, de-ess, manually limit , among other surgical fixes here and there. It's still far from 'good'. Usable? That's subjective.. https://www.dropbox.com/s/tpuizuuu1w...essed.mp3?dl=0

March 4th, 2016, 01:06 PM	#6
Bernie Beaudry Regular Crew Join Date: Oct 2007 Location: St. Paul, Minnesota Posts: 166	Re: Another attempt to fix this awful file Hi Kathy, I've been learning RX 4 so I took your file on as a challenge. I got rid of much of the white noise which uncovered some artifacts in your audio. It sounds like the back ground was getting hit with some sort of auto gain or something. I hear bits of stuff but with music I don't think you'd hear it. I did two passes of Spectral Denoiser, then eq, another pass of Spectral Denoiser, then Declip. I then eq'd one of the sections that had a duller sound than the rest. Here you go. Tell me what you think. https://www.dropbox.com/s/rmqeh3aiv4...20fix.wav?dl=0

March 4th, 2016, 05:19 PM	#10
Bernie Beaudry Regular Crew Join Date: Oct 2007 Location: St. Paul, Minnesota Posts: 166	Re: Another attempt to fix this awful file Here's another version: https://www.dropbox.com/s/xl2cym5jp2..._Bad3.wav?dl=0 The problem with this one is the voice still is getting a bit robotized.

March 5th, 2016, 03:11 AM	#11
Christopher Young Trustee Join Date: Apr 2008 Location: Sydney Australia Posts: 1,528	Re: Another attempt to fix this awful file Yes a problematic little file indeed. It has various wide band artifacts spread throughout its spectrum, I also think there is a bit of flanging going on in there some how? How come who knows. As previously mentioned when you try to clean it up it gets a bit robotic. I found the best I could get out of it was to attack it pretty hard on NR then give it a bit of normalizing compressing and limiting. Finally I added back -42dB of White Noise as it was sounding too dry. Best I could do with in the time to hand. https://www.sendspace.com/pro/dl/nikklv

March 5th, 2016, 09:06 AM	#15
Greg Miller Trustee Join Date: Jan 2011 Location: Pennsylvania Posts: 1,786	Re: Another attempt to fix this awful file Sorry for the confusion about the hum (or non-hum). I was just looking for an explanation for the strange vocal distortion that I seem to hear in several spots. How sad that there is no earlier generation or better version of this file. :-( Out of curiosity, was it recorded with a wired mic, or with wireless? I hear a little noise burst, just after "... all the way around it" and I wondered whether that was an RF hit. Or maybe it's a clothing rub ... the white noise really obscures what's going on there.