Perfect Declipper

[Dogen]

Another issue with many current recordings is that it’s basically mono, I guess because so many listeners use mono Bluetooth speakers. It’s hard to find new recordings that intelligently use two channels to create any kind of separation. Has anyone measured to see if this is actually the case?

 

[dallasjustice]

I haven’t been using perfect declipper. Roon is my preferred playback source so I can’t use .vst plugins with Roon. I liked what perfect declipper did but it’s not worth it for me to have to use Jriver instead of Roon.

 

[tmtomh]

Very interesting how long does it takes to do an conversion/declipping lets say per minute.

Depends on the software, how powerful your computer is, and where you set the de-clipping threshold (takes longer if the software has to reconstruct a ton of samples versus a relatively smaller number of peaks). That said, on my old 2009 iMac using Izotope, it typically takes about a minute to scan the waveform (not strictly necessary, but very helpful because it allows the software to suggest a de-clipping threshold - although the suggestion only sometimes is right). Then it takes about 3 minutes to de-clip and album, and another minute to re-scan the new waveform. So the actual de-clipping I would estimate at about 4-5 seconds per minute of music, but for the entire process - including scanning, re-scanning and exporting/saving the de-clipped file - I would estimate more like 8 seconds per minute of music.

Also, keep in mind that de-clipping really works best - and quickest - if you run it on the entire album as a single file. So that means you have to allot a few extra minutes before and after de-clipping for (A) combining your tracks into a single album-length file, and (B) splitting the de-clipped result back into tracks.

 

[Snarfie]

OK i tried out freeware Audacity using the Clip fix tool using Adel Album the horrible compressed 21.
Settings for clip fix where mostly threshold setting at least 80% using some tracks (about 3) didn't need that so threshold varies between 80 till 10%

Enclosed Original Dr and Audacity Dr measurement. I got it from 7Dr original to 12Dr Audacity. When listening yes there is a difference voice bit more clear is not drowned anymore between instruments. Bass bit more tight. Overal it sound more plesant for your ears less fatigue. But i did aspect a larger difference. Will Perfect Declipper do a better job?. What is the difference between Perfect Declipper & Audacity regarding decompressing music.

foobar2000 1.4.1 / Dynamic Range Meter 1.1.1
log date: 2019-01-01 17:22:50

--------------------------------------------------------------------------------
Analyzed: Adele / 21
--------------------------------------------------------------------------------

DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR5 -0.02 dB -7.81 dB 3:49 01/13-Rolling In The Deep
DR5 -0.02 dB -6.78 dB 3:41 02/13-Rumor Has It
DR6 -0.01 dB -8.73 dB 4:12 03/13-Turning Tabels
DR6 -0.02 dB -8.88 dB 4:03 04/13-Dont You Remember
DR5 -0.01 dB -6.88 dB 4:02 05/13-Set Fire To The Rain
DR7 -0.02 dB -9.62 dB 4:38 06/13-He Wont Go
DR8 -0.01 dB -11.14 dB 3:48 07/13-Take It All
DR5 -0.01 dB -6.39 dB 4:02 08/13-Ill Be Waiting
DR7 -0.02 dB -9.79 dB 5:47 09/13-One And Only
DR7 -0.02 dB -9.82 dB 5:18 10/13-Lovesong
DR8 -0.02 dB -10.40 dB 4:45 11/13-Someone Like You
DR8 -0.09 dB -9.56 dB 3:08 12/13-If It Handnt Been For Love
DR8 -0.10 dB -10.44 dB 3:30 13/13-Hiding My Heart
--------------------------------------------------------------------------------

Number of tracks: 13
Official DR value: DR7

Samplerate: 44100 Hz
Channels: 2
Bits per sample: 16
Bitrate: 1411 kbps
Codec: PCM
================================================================================


foobar2000 1.4.1 / Dynamic Range Meter 1.1.1
log date: 2019-01-01 21:48:55

--------------------------------------------------------------------------------
Analyzed: Adele / 21
--------------------------------------------------------------------------------

DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR11 0.00 dB -13.86 dB 3:49 01-Rolling In The Deep
DR11 0.00 dB -12.83 dB 3:41 02-Rumor Has It
DR10 -0.25 dB -14.72 dB 4:12 03-Turning Tabels
DR12 0.00 dB -14.93 dB 4:03 04-Dont You Remember
DR11 0.00 dB -12.90 dB 4:02 05-Set Fire To The Rain
DR13 0.00 dB -15.69 dB 4:38 06-He Wont Go
DR14 0.00 dB -16.95 dB 3:48 07-Take It All
DR11 0.00 dB -12.51 dB 4:02 08-Ill Be Waiting
DR13 0.00 dB -15.86 dB 5:47 09-One And Only
DR13 0.00 dB -15.83 dB 5:18 10-Lovesong
DR13 0.00 dB -15.86 dB 4:45 11-Someone Like You
DR14 0.00 dB -15.64 dB 3:08 12-If It Handnt Been For Love
DR14 0.00 dB -16.07 dB 3:30 13-Hiding My Heart
--------------------------------------------------------------------------------

Number of tracks: 13
Official DR value: DR12

Samplerate: 44100 Hz
Channels: 2
Bits per sample: 16
Bitrate: 1411 kbps
Codec: PCM
================================================================================

 

[pos]

I did a little experiment with this declipper yesterday, and posted 6 seconds clips here: https://www.audiosciencereview.com/...ds/thimeo-perfect-declipper.5966/#post-133972
The difference is pretty audible

 

[solderdude]

wouldn't it be fun to take a not clipped recording.
Add some gain so it clips here and there and then run it through a de-clipper.
Then adjust the amplitude of the de-clipped file so it matches the original again and null.
After that analyze those nulls to see how well lost info is retrieved ?

 

[tmtomh]

OK i tried out freeware Audacity using the Clip fix tool using Adel Album the horrible compressed 21.
Settings for clip fix where mostly threshold setting at least 80% using some tracks (about 3) didn't need that so threshold varies between 80 till 10%

Enclosed Original Dr and Audacity Dr measurement. I got it from 7Dr original to 12Dr Audacity. When listening yes there is a difference voice bit more clear is not drowned anymore between instruments. Bass bit more tight. Overal it sound more plesant for your ears less fatigue. But i did aspect a larger difference. Will Perfect Declipper do a better job?. What is the difference between Perfect Declipper & Audacity regarding decompressing music.

foobar2000 1.4.1 / Dynamic Range Meter 1.1.1
log date: 2019-01-01 17:22:50

--------------------------------------------------------------------------------
Analyzed: Adele / 21
--------------------------------------------------------------------------------

DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR5 -0.02 dB -7.81 dB 3:49 01/13-Rolling In The Deep
DR5 -0.02 dB -6.78 dB 3:41 02/13-Rumor Has It
DR6 -0.01 dB -8.73 dB 4:12 03/13-Turning Tabels
DR6 -0.02 dB -8.88 dB 4:03 04/13-Dont You Remember
DR5 -0.01 dB -6.88 dB 4:02 05/13-Set Fire To The Rain
DR7 -0.02 dB -9.62 dB 4:38 06/13-He Wont Go
DR8 -0.01 dB -11.14 dB 3:48 07/13-Take It All
DR5 -0.01 dB -6.39 dB 4:02 08/13-Ill Be Waiting
DR7 -0.02 dB -9.79 dB 5:47 09/13-One And Only
DR7 -0.02 dB -9.82 dB 5:18 10/13-Lovesong
DR8 -0.02 dB -10.40 dB 4:45 11/13-Someone Like You
DR8 -0.09 dB -9.56 dB 3:08 12/13-If It Handnt Been For Love
DR8 -0.10 dB -10.44 dB 3:30 13/13-Hiding My Heart
--------------------------------------------------------------------------------

Number of tracks: 13
Official DR value: DR7

Samplerate: 44100 Hz
Channels: 2
Bits per sample: 16
Bitrate: 1411 kbps
Codec: PCM
================================================================================


foobar2000 1.4.1 / Dynamic Range Meter 1.1.1
log date: 2019-01-01 21:48:55

--------------------------------------------------------------------------------
Analyzed: Adele / 21
--------------------------------------------------------------------------------

DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR11 0.00 dB -13.86 dB 3:49 01-Rolling In The Deep
DR11 0.00 dB -12.83 dB 3:41 02-Rumor Has It
DR10 -0.25 dB -14.72 dB 4:12 03-Turning Tabels
DR12 0.00 dB -14.93 dB 4:03 04-Dont You Remember
DR11 0.00 dB -12.90 dB 4:02 05-Set Fire To The Rain
DR13 0.00 dB -15.69 dB 4:38 06-He Wont Go
DR14 0.00 dB -16.95 dB 3:48 07-Take It All
DR11 0.00 dB -12.51 dB 4:02 08-Ill Be Waiting
DR13 0.00 dB -15.86 dB 5:47 09-One And Only
DR13 0.00 dB -15.83 dB 5:18 10-Lovesong
DR13 0.00 dB -15.86 dB 4:45 11-Someone Like You
DR14 0.00 dB -15.64 dB 3:08 12-If It Handnt Been For Love
DR14 0.00 dB -16.07 dB 3:30 13-Hiding My Heart
--------------------------------------------------------------------------------

Number of tracks: 13
Official DR value: DR12

Samplerate: 44100 Hz
Channels: 2
Bits per sample: 16
Bitrate: 1411 kbps
Codec: PCM
================================================================================

DR 7 to DR 12 is a major difference - what "larger difference" were you expecting? Or are you saying you were expecting a larger difference in the sonics when you listened to it? Honestly, a good number of overly compressed albums are not actually badly mastered other than the compression - often the EQ is about right, the clarity is good, and the recording quality is decent. (There are of course many horribly over-produced, auto-tuned-to-death productions too; but I don't think the Adele album is one of them.)

So if the de-clipped version sounds similar, but less fatiguing and bit more open, I'd say that's a very good thing. The goal of de-clipping is not usually to make it sound like a totally different mix, just to undo a good bit of the digital limiting that was applied at (usually) the final post-mix mastering stage.

 

[Snarfie]

DR 7 to DR 12 is a major difference - what "larger difference" were you expecting? Or are you saying you were expecting a larger difference in the sonics when you listened to it? Honestly, a good number of overly compressed albums are not actually badly mastered other than the compression - often the EQ is about right, the clarity is good, and the recording quality is decent. (There are of course many horribly over-produced, auto-tuned-to-death productions too; but I don't think the Adele album is one of them.)

So if the de-clipped version sounds similar, but less fatiguing and bit more open, I'd say that's a very good thing. The goal of de-clipping is not usually to make it sound like a totally different mix, just to undo a good bit of the digital limiting that was applied at (usually) the final post-mix mastering stage.

Yes the Dr rating is greatly improved but from a sound perspective the difference is not what i expected depth seperation etc. (as you already mentiond don't expect a totally different mix) but for sure it sounds better (not great) because it is less fatigue an voices instruments are a bit more separated. So the Audacity declipping is the one i listen too not the original anymore.

It is known that the Adel CD albums don't sound good on decent audio gear. Only in the car it shoud sounds good. Have a look at Dr loudness wars http://dr.loudness-war.info/album/list?artist=Adele&album=25 or Google on Adel albums "badly mastered". For sure the people that recorded the original master tapes know what they did. But on behave of the record company they commercialy mixed it down such that it has to sound good on a car stereo. I really hope that one day they take the original master tape an mix it such that it sounds good on descent audio gear think ther is hope because the Vinyl version has a way higher Dr rating than the CD. The same can be said from Amy Winhouse however with Back to Black one can argue the producers wanted a 60tie sound and here voice an instruments went threw a digitalizations proces to create an desired atmosphere including overall lots of compression so partly artisticly it sound as it is.