Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Help me understand the bitcompare track results. (Read 5231 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Help me understand the bitcompare track results.

Hi Friends

I have done two bitcompare track procedures using foobar2000 bitcompare. Could you please help me understand the results properly

Scenario 1:- Same track in same movie but two different editions of CDs

Here is the results

Code: [Select]
Comparing:
"Tell Me Why_edition1.flac"
"Tell Me Why_edition2.flac"
Compared 11340756 samples.
Differences found: 22413039 values, starting at 0:00.295465, peak: 1.2105713 at 0:32.057347, 1ch
Channel difference peaks: 1.2105713 1.0881653
File #1 peaks: 0.8833923 0.7939758
File #2 peaks: 0.8833923 0.7939758
Detected offset as 730 samples.

Comparing again with corrected offset...
Compared 11340026 samples, with offset of 730 discarding last/first samples from total of 11340756, discarded samples were silent in both files.
Differences found within the compared range: 61 values, starting at 0:26.814399, peak: 0.0386658 at 0:43.222517, 1ch
Channel difference peaks: 0.0386658 0.0244446
File #1 peaks: 0.8833923 0.7939758
File #2 peaks: 0.8833923 0.7939758

As far as I understand both tracks have same number of samples and same peak levels, But what is "Differences found within the compared range: 61 values, starting at 0:26.814399, peak: 0.0386658 at 0:43.222517, 1ch
Channel difference peaks: 0.0386658 0.0244446" ?

Scenario 2:- Normal song in CD1 and same song remastered in CD2

Code: [Select]
Differences found in compared tracks.

Comparing:
"track1.flac"
"track1_remastered.flac"
Differences found: length mismatch - 4:12.786667 vs 4:14.813333, 11147892 vs 11237268 samples.
Compared 11147892 samples, discarded last 89376 samples from the longer file.
Differences found within the compared range: 22148453 values, starting at 0:00.613288, peak: 1.2126770 at 0:18.514558, 1ch
Channel difference peaks: 1.2126770 1.0762939
File #1 peaks: 0.8758545 0.9999695
File #2 peaks: 0.6316528 0.6376953



Total duration processed: 4:12.787
Time elapsed: 0:03.471
72.82x realtime


I see there are more samples in the remastered track but it has lower peaks. Is there anything more to this?

Cheers

Re: Help me understand the bitcompare track results.

Reply #1
As far as I understand both tracks have same number of samples and same peak levels, But what is "Differences found within the compared range: 61 values, starting at 0:26.814399, peak: 0.0386658 at 0:43.222517, 1ch
Channel difference peaks: 0.0386658 0.0244446" ?

There are some minor differences in a few samples, possibly clipping. 

Scenario 2:- Normal song in CD1 and same song remastered in CD2

I see there are more samples in the remastered track but it has lower peaks. Is there anything more to this?

Peak is higher, not lower.  The files are from a different mastering, with different file lengths, and different contents.  The tool is telling you that everything is different.

Re: Help me understand the bitcompare track results.

Reply #2
There are some minor differences in a few samples, possibly clipping. 

Possibly interpolation, you mean?

The explanation could be errors in ripping the CDs, but also errors in the manufactoring process. (What if one of the pressing plants was sent a CD in the mail to master from, and ...)

Re: Help me understand the bitcompare track results.

Reply #3
As far as I understand both tracks have same number of samples and same peak levels, But what is "Differences found within the compared range: 61 values, starting at 0:26.814399, peak: 0.0386658 at 0:43.222517, 1ch
Channel difference peaks: 0.0386658 0.0244446" ?

There are some minor differences in a few samples, possibly clipping. 

Scenario 2:- Normal song in CD1 and same song remastered in CD2

I see there are more samples in the remastered track but it has lower peaks. Is there anything more to this?

Peak is higher, not lower.  The files are from a different mastering, with different file lengths, and different contents.  The tool is telling you that everything is different.

What does "Differences found within the compared range: 61 values, starting at 0:26.814399, peak: 0.0386658 at 0:43.222517, 1ch" mean?

Regarding the peaks

File #1 peaks: 0.8758545 0.9999695
File #2 peaks: 0.6316528 0.6376953  --->remastered track

Remastered has low peaks isnt it?

Re: Help me understand the bitcompare track results.

Reply #4
So fb2k has first found that the music is shifted 730 samples in between the two. After shifting by 730 samples and cropping down to those which exist in both signals:
What does "Differences found within the compared range: 61 values, starting at 0:26.814399, peak: 0.0386658 at 0:43.222517, 1ch" mean?
... then there are 61 samples where the files differ. The first is at 0:26.814399. The largest difference is 0.0386658 and occurs in channel 1 (left) at time 0:43.222517.

0.0386658 means it is about 14 dB below digital full volume.

Do you have both the physical CDs and rip logs?

Re: Help me understand the bitcompare track results.

Reply #5
So fb2k has first found that the music is shifted 730 samples in between the two. After shifting by 730 samples and cropping down to those which exist in both signals:
What does "Differences found within the compared range: 61 values, starting at 0:26.814399, peak: 0.0386658 at 0:43.222517, 1ch" mean?
... then there are 61 samples where the files differ. The first is at 0:26.814399. The largest difference is 0.0386658 and occurs in channel 1 (left) at time 0:43.222517.

0.0386658 means it is about 14 dB below digital full volume.

Do you have both the physical CDs and rip logs?


Yes I do have the logs


Regarding the peaks, what do they indicate. How is it related to audible sound?

File #1 peaks: 0.8758545 0.9999695
File #2 peaks: 0.6316528 0.6376953  --->remastered track

Remastered has low peaks isnt it?

Re: Help me understand the bitcompare track results.

Reply #6
Yes I do have the logs
Anything worth reporting?

What does it say if you try to verify with CUETools?


File #1 peaks: 0.8758545 0.9999695
File #2 peaks: 0.6316528 0.6376953  --->remastered track

Remastered has low peaks isnt it?
Yes. Convert to dB: http://www.playdotsound.com/portfolio-item/decibel-db-to-float-value-calculator-making-sense-of-linear-values-in-audio-tools/
(And I typed wrong above.)

Re: Help me understand the bitcompare track results.

Reply #7
Yes I do have the logs
Anything worth reporting?

What does it say if you try to verify with CUETools?


File #1 peaks: 0.8758545 0.9999695
File #2 peaks: 0.6316528 0.6376953  --->remastered track

Remastered has low peaks isnt it?
Yes. Convert to dB: http://www.playdotsound.com/portfolio-item/decibel-db-to-float-value-calculator-making-sense-of-linear-values-in-audio-tools/
(And I typed wrong above.)

Sorry dont have much idea about CUETools. But here is the verify results
[CUETools log; Date: 24/02/2018 12:32:18 AM; Version: 2.1.5]
[CTDB TOCID: muhGQ8LRpTfvYelQ_qz1XbnCHB0-] found.
Track | CTDB Status
  1   | (3/4) Accurately ripped
  2   | (4/4) Accurately ripped
  3   | (4/4) Accurately ripped
  4   | (4/4) Accurately ripped
  5   | (4/4) Accurately ripped
  6   | (4/4) Accurately ripped
  7   | (3/4) Accurately ripped
  8   | (4/4) Accurately ripped
  9   | (3/4) Accurately ripped
 10   | (4/4) Accurately ripped
 11   | (4/4) Accurately ripped
 12   | (4/4) Accurately ripped
 13   | (4/4) Accurately ripped
 14   | (4/4) Accurately ripped
 15   | (3/4) Accurately ripped
[AccurateRip ID: 0022f593-018d45a2-e70ebc0f] found.
Track   [  CRC   |   V2   ] Status
 01     [48399aee|e3e19126] (0+0/1) No match
 02     [a9ac5d6f|bb100e6b] (0+0/1) No match
 03     [2e325d23|b2ad8816] (0+0/1) No match
 04     [a6b20608|71a93f10] (0+0/1) No match
 05     [3e1511ba|b439176f] (0+0/1) No match
 06     [2c4d37c7|4200fbd1] (0+0/1) No match
 07     [7172b9a6|4213d068] (0+0/1) No match
 08     [8168b351|35386004] (0+0/1) No match
 09     [45d9a381|73d2197b] (0+0/1) No match
 10     [d5e568c1|6e5fec52] (0+0/1) No match
 11     [63ea6118|acf4cd25] (0+0/1) No match
 12     [3f0f7633|f7fe85a1] (0+0/1) No match
 13     [d3c64de4|c4ff6b43] (0+0/1) No match
 14     [fd6b6a91|f73f9dba] (0+0/1) No match
 15     [a4737309|1dc1b029] (0+0/1) No match
Offsetted by 730:
 01     [5076bfea] (0/1) No match (V2 was not tested)
 02     [31ce4601] (0/1) No match (V2 was not tested)
 03     [582b10eb] (0/1) No match (V2 was not tested)
 04     [0de93ba2] (0/1) No match (V2 was not tested)
 05     [af9fd2b4] (0/1) No match (V2 was not tested)
 06     [41ab855f] (0/1) No match (V2 was not tested)
 07     [4b005c56] (0/1) No match (V2 was not tested)
 08     [8d6baafb] (0/1) No match (V2 was not tested)
 09     [51dbe99f] (0/1) No match (V2 was not tested)
 10     [ea24cf71] (0/1) No match (V2 was not tested)
 11     [d762061a] (0/1) No match (V2 was not tested)
 12     [3259360d] (0/1) No match (V2 was not tested)
 13     [650b1b68] (0/1) No match (V2 was not tested)
 14     [84f0b05f] (0/1) No match (V2 was not tested)
 15     [818e10ac] (0/1) No match (V2 was not tested)

Track Peak [ CRC32  ] [W/O NULL] [  LOG   ]
 --   96.3 [E25372C3] [D854B2B0]          
 01   88.3 [224524CA] [CA1F09C9]   CRC32  
 02   88.0 [55E0EC9E] [F2262036]   CRC32  
 03   85.9 [DFEA8DBE] [CD736974]   CRC32  
 04   95.2 [5B4D005A] [B82F8EAF]   CRC32  
 05   87.3 [2FF9B1DF] [C400B8EA]   CRC32  
 06   86.6 [236BBF2D] [9543E7F6]   CRC32  
 07   96.3 [C0E99FF7] [5C25D44C]   CRC32  
 08   78.9 [D7F9602D] [332AB14A]   CRC32  
 09   87.5 [F3E9EBC2] [76F23DCC]   CRC32  
 10   86.1 [AB28C754] [34E23B9B]   CRC32  
 11   89.6 [A7141BF6] [6404C821]   CRC32  
 12   80.9 [62C8C4E4] [D0C053A0]   CRC32  
 13   86.0 [D3887216] [4228E438]   CRC32  
 14   84.3 [AB0E41B5] [DEE2A351]   CRC32  
 15   90.2 [C9E6580C] [705CA9AF]   CRC32  


I used the converter and got the results.
File #1 peaks: 0.8758545(-1.1513606870375876 dB) 0.9999695(-0.000264923674067936 dB)
File #2 peaks: 0.6316528(-3.99043148745395 dB) 0.6376953(-3.907735677554161 dB)  --->remastered track

Can you please through some light what this means (peak in linear and dB) in terms of sound quality?

 

Re: Help me understand the bitcompare track results.

Reply #8
Can you please through some light what this means (peak in linear and dB) in terms of sound quality?
Basically, nothing.


Re: Help me understand the bitcompare track results.

Reply #10
It's just little extra information. In your example it shows that remastered track's loudest single sample is about 4 dB quieter than in the other file. It can't tell how loud the files are though, if the remaster is heavily compressed it can still be louder than the original.

If one file had peak at 0.0 it would mean the file is nothing but silence.

If the peaks were close to each other and the highest differences were small, it would mean the files are very similar. Possibly audibly identical.

The peak info can also show clipping. Psychoacoustic lossy formats like MP3 and lossless formats that support floating point can exceed the digital fullscale's 1.0 value. If such a file is compressed to regular PCM the peaks would get clipped to 1.0. Bitcompare would show this.

Re: Help me understand the bitcompare track results.

Reply #11
It's just little extra information. In your example it shows that remastered track's loudest single sample is about 4 dB quieter than in the other file. It can't tell how loud the files are though, if the remaster is heavily compressed it can still be louder than the original.

If one file had peak at 0.0 it would mean the file is nothing but silence.

If the peaks were close to each other and the highest differences were small, it would mean the files are very similar. Possibly audibly identical.

The peak info can also show clipping. Psychoacoustic lossy formats like MP3 and lossless formats that support floating point can exceed the digital fullscale's 1.0 value. If such a file is compressed to regular PCM the peaks would get clipped to 1.0. Bitcompare would show this.


I have a few more questions.

"Differences found: 22413039 values, starting at 0:00.295465, peak: 1.2105713 at 0:32.057347, 1ch"

In the differences found the highest peak is at 1.2105713. What does this mean. As you said above 1 means clipping?


Also "Channel difference peaks: 0.0386658 0.0244446"

What does channel difference peaks mean.

what is the dB value for lossless formats

Re: Help me understand the bitcompare track results.

Reply #12
In the differences found the highest peak is at 1.2105713. What does this mean. As you said above 1 means clipping?
Highest difference peak means the largest difference between sample values in the two files at certain position. In this case the spot where the highest difference was found was a little after 32 seconds from the start. In floating point format the samples are normally between -1.0 and +1.0 so the highest difference without either file exceeding the scale can be 2.0. In your case one file could have for example had sample value 1.0 and the other file -0.2105713. The difference between these two values is 1.2105713.

Also "Channel difference peaks: 0.0386658 0.0244446"

What does channel difference peaks mean.
CD audio is stereo and has two channels. The comparison tool shows the highest difference between each channel separately. First value is for the left channel and the second value is highest difference in the right channel.

You may notice that the differences here are quite tiny. Converted to decibels the left channel difference is 0.34 dB and the right is 0.21 dB. And as these are from the mostly identical tracks with only 61 samples being different (that's 1 millisecond of total difference in the entire track) it's quite safe to say no one can hear any difference between these two files.

what is the dB value for lossless formats
Not sure what you mean. These tiny differences can be converted to decibels but I don't think such conversion applies to all track pairs or when differences exceed 1.0. As an example exact same track compared against itself with inverted polarity can show peak differences at 2.0 (or higher if it's lossy or float) but the two  files will sound exactly identical.

Re: Help me understand the bitcompare track results.

Reply #13
My, if I had a $ for each time I've made a mistake from forgetting that one bit is the sign, then I would probably have ... some positive dollars and some negative dollars, so many that the (normalized) difference would close in on the normal distribution.

Re: Help me understand the bitcompare track results.

Reply #14
In the differences found the highest peak is at 1.2105713. What does this mean. As you said above 1 means clipping?
Highest difference peak means the largest difference between sample values in the two files at certain position. In this case the spot where the highest difference was found was a little after 32 seconds from the start. In floating point format the samples are normally between -1.0 and +1.0 so the highest difference without either file exceeding the scale can be 2.0. In your case one file could have for example had sample value 1.0 and the other file -0.2105713. The difference between these two values is 1.2105713.

Also "Channel difference peaks: 0.0386658 0.0244446"

What does channel difference peaks mean.
CD audio is stereo and has two channels. The comparison tool shows the highest difference between each channel separately. First value is for the left channel and the second value is highest difference in the right channel.

You may notice that the differences here are quite tiny. Converted to decibels the left channel difference is 0.34 dB and the right is 0.21 dB. And as these are from the mostly identical tracks with only 61 samples being different (that's 1 millisecond of total difference in the entire track) it's quite safe to say no one can hear any difference between these two files.

what is the dB value for lossless formats
Not sure what you mean. These tiny differences can be converted to decibels but I don't think such conversion applies to all track pairs or when differences exceed 1.0. As an example exact same track compared against itself with inverted polarity can show peak differences at 2.0 (or higher if it's lossy or float) but the two  files will sound exactly identical.

Converted to decibels the left channel difference is 0.34 dB and the right is 0.21 dB --> How did you convert this to dB. The link posted above shows a different value after conversion.

Regarding db. For lossless music files what the dB level of max audible capacity. (20 - 20Khz for human audible frequency).

Re: Help me understand the bitcompare track results.

Reply #15
How did you convert this to dB. The link posted above shows a different value after conversion.
I moved the reference to digitall full scale. 20*log(1.0 - difference).

If you compare directly computed dB values you don't get the idea how small the loudness differences are. As an example let's take two sample values: 0.000001 (-120 dB) and 0.000002 (-114 dB). Comparing their loudness directly shows that one is twice as loud so 6 dB difference, but in reality both samples are completely inaudible in normal playback. When the difference is moved to reference point the dB difference (20*log(1-0.000001)) is about 0.000009 dB.

Regarding db. For lossless music files what the dB level of max audible capacity. (20 - 20Khz for human audible frequency).
If you mean the signal to noise ratio of different bitdepths the formula is 20*log(2^bits). 16-bit audio shows 96 dB SNR and 24-bit audio 144 dB SNR.

Best audio equipment has SNR of about 120 dB. Common equipment about 100 dB.