Skip to main content

Notice

If you are using a Hotmail or Outlook email address, please change it now, as Microsoft is rejecting all email from our service outright.
Topic: How to determine which is better by seeing waveform? (Read 1918 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

How to determine which is better by seeing waveform?

These are same song from different CD albums

One is from 1995 year which is first release
Another one is from 2003 year

Which one is better?
Why?


X

Re: How to determine which is better by seeing waveform?

Reply #1
This image is from Audacity, right?  Turn on View > Show Clipping.

Re: How to determine which is better by seeing waveform?

Reply #2
Usually older releases up to year 1993-1995 are better quality with better dynamics. But you should level match them (to similar average loudness), for example with replaygain, and compare by listening. It might be that the older copy is lacking in bandwidth, equalization or noise processing, and the slightly more compressed version is overall preferable.

Re: How to determine which is better by seeing waveform?

Reply #3
you also want to run them through sox (as a great example) to give you a spectral analysis, thereby giving you some more info to analyse.

Install sox, then copy & paste the following into a file called spectrograms.cmd and save into the same folder as that which you just installed sox into (NOT C:\Program Files / C:\Program Files (x86)):
Code: [Select]
c:
cd %~dp0
mkdir Spectrograms
FOR %%A IN (%*) DO sox %%A -n remix 1 spectrogram -x 3000 -y 513 -z 120 -w Kaiser -o "Spectrograms\%%~nxA-full.png"
FOR %%A IN (%*) DO sox %%A -n remix 1 spectrogram -X 500 -y 1025 -z 120 -w Kaiser -S 1:00 -d 0:02 -o "Spectrograms\%%~nxA-zoom.png"
pause

Then drag and drop your flac or wav file onto the spectrograms.cmd and what for the magic to happen.

Re: How to determine which is better by seeing waveform?

Reply #4
Quote
Which one is better?
Whichever one you prefer!    L:isten with your ears...   You can't judge sound quality with waveforms, and charts & graphs.  (But you might want to match the volumes before A/B listening to decide which one you prefer.)

Quote
Why?
Because it's your music for your enjoyment. ;)

The newer one is probably a victim (or victor) in The Loudness War.

Most music lovers like more dynamics and they'd probably prefer the older less compressed version, assuming no other differences.    But if it was loudness-ware remastering somebody thought the newer-louder version was "better".




Re: How to determine which is better by seeing waveform?

Reply #5
you also want to run them through sox (as a great example) to give you a spectral analysis, thereby giving you some more info to analyse.
[snip]
Then drag and drop your flac or wav file onto the spectrograms.cmd and what for the magic to happen.
or just drag n drop the files into spek: http://spek.cc/

foobar's abx comparator is also a great tool for comparing different masters

Re: How to determine which is better by seeing waveform?

Reply #6
SoX is better for spectrograms. It allows to select a range of time and amplitude for zooming in, such as the first and last seconds of a recording, resample inputs to the same rate for a more intuitive comparison and present the difference between stereo channels. Large spectrograms gan be generated in batch mode and then quickly examined. And the command-line for those parameters can be saved as a preset. Spek requires to use keyboard shortcuts to adjust the dB floor, does not remember the setting, and there is no way to trim or zoom into a time interval.

I like to routinely look at spectrograms from 10 seconds around track boundaries to see if cue points are accurate and if endings are cut off with noise reduction. A CD could also be better than another if the cues match the start of music on one but not the other.

Re: How to determine which is better by seeing waveform?

Reply #7
The first track has an uncompressed waveform and a good dynamic range.
The second track has compressed and close to the uniform waveform, which means degraded dynamic range sound like musical soup with no detailed sound of any musical instruments with boosted loudness as result of the pointless and worthless loudness war.

The best explanation of what is loudness war  https://www.youtube.com/watch?v=dcKDMBuGodU

Re: How to determine which is better by seeing waveform?

Reply #8
SoX is better for spectrograms. It allows to select a range of time and amplitude for zooming in, such as the first and last seconds of a recording, resample inputs to the same rate for a more intuitive comparison and present the difference between stereo channels. Large spectrograms gan be generated in batch mode and then quickly examined. And the command-line for those parameters can be saved as a preset. Spek requires to use keyboard shortcuts to adjust the dB floor, does not remember the setting, and there is no way to trim or zoom into a time interval.

I like to routinely look at spectrograms from 10 seconds around track boundaries to see if cue points are accurate and if endings are cut off with noise reduction. A CD could also be better than another if the cues match the start of music on one but not the other.
Exactly.

The .cmd (or .bat if one prefers) that I suggested is created in my post above is quick and easy to use and modify if need be.

Re: How to determine which is better by seeing waveform?

Reply #9
Code: [Select]
c:
cd %~dp0
mkdir Spectrograms
FOR %%A IN (%*) DO sox %%A -n remix 1 spectrogram -x 3000 -y 513 -z 120 -w Kaiser -o "Spectrograms\%%~nxA-full.png"
FOR %%A IN (%*) DO sox %%A -n remix 1 spectrogram -X 500 -y 1025 -z 120 -w Kaiser -S 1:00 -d 0:02 -o "Spectrograms\%%~nxA-zoom.png"
pause
Not sure if that's intentional but "remix 1" will pick only the left channel. You need either "remix 1-2" or "channels 1" if you want them mixed (or nothing if you want separate spectrograms per channel):
Code: [Select]
sox -r44.1k -c2 -n -b24 L1k-R5k.wav synth 5 sin 1k sin 5k fade 0.2 0 0.2 norm -6
sox L1k-R5k.wav -n remix 1    spectrogram -x 1136 -o remix1.png
sox L1k-R5k.wav -n remix 1-2  spectrogram -x 1136 -o remix12.png
sox L1k-R5k.wav -n channels 1 spectrogram -x 1136 -o channels1.png

Re: How to determine which is better by seeing waveform?

Reply #10
This image is from Audacity, right?  Turn on View > Show Clipping.

Is that mean 2003 version better?

But I don't hear any problem/clipping in 1995 version

X

Re: How to determine which is better by seeing waveform?

Reply #11
you also want to run them through sox (as a great example) to give you a spectral analysis, thereby giving you some more info to analyse.

Install sox, then copy & paste the following into a file called spectrograms.cmd and save into the same folder as that which you just installed sox into (NOT C:\Program Files / C:\Program Files (x86)):
Code: [Select]
c:
cd %~dp0
mkdir Spectrograms
FOR %%A IN (%*) DO sox %%A -n remix 1 spectrogram -x 3000 -y 513 -z 120 -w Kaiser -o "Spectrograms\%%~nxA-full.png"
FOR %%A IN (%*) DO sox %%A -n remix 1 spectrogram -X 500 -y 1025 -z 120 -w Kaiser -S 1:00 -d 0:02 -o "Spectrograms\%%~nxA-zoom.png"
pause

Then drag and drop your flac or wav file onto the spectrograms.cmd and what for the magic to happen.



What can I see from these spectrograms?

1995 full:
X

2003 full:
X

1995 zoom:
X

2003 zoom:
X

Re: How to determine which is better by seeing waveform?

Reply #12
imo...

- both show lossless files (which makes sense since you say they're lossless CD rips).
- in the 2003 there is a cutoff of sorts at ~21.5kHz. Presumably imposed upon it during the (re)mastering of this audio for the 2003 release. Given the "loudness" / level increase (as shown from your Audacity waveform images) then perhaps the audio frequency range has been affected due to the increase in levels and likely compression, which is part of the "loudness war" that DVDdoug mentioned.

Do you know what source they used for the 2003 release? Was it master tapes?
It's quite possible they simply used the CD audio from the 1995 release and that's why it shows signs of a lowered frequency range after processing/mastering etc. Wouldn't be smart to do that mind you, but it's not unheard of.

I would suggest Replaygaining both lossless releases and listening to them, to see what you think? Can you tell a difference? If so, is one preferable to the other? j7n also said to do this and I agree.

That being said, based on just the info you've provided in this thread I would go with the 1995 release. My experience of early-mid 90s releases and earlier, compared to 2000+ remasters, is that the early ones tend to have better dynamics (from spectral analysis). But again, once Replaygained can a difference be heard? For those with good hearing, maybe.


 
SimplePortal 1.0.0 RC1 © 2008-2021