Is it possible to convert an mp3 file with a lower bitrate to a higher one?
For example, I have an mp3 wich is encoded at 192 CBR, is it possible for me to convert it to 320 kbps?

Because I think I've just downloaded a fake 320kbps discography

Can someone please take a look at this ant tell me whether it's real 320 mp3s or not??? (When I play it, it says 320, but I don't thinks so because of the sound quality)




EAC extraction logfile from 10. November 2006, 20:37 for CD
Good Charlotte / The Chronicles of Life and Death (DEATH Version)

Used drive : Optiarc DVD+-RW ND-3570A Adapter: 0 ID: 0
Read mode : Secure with NO C2, accurate stream, disable cache
Read offset correction : 48
Overread into Lead-In and Lead-Out : No

Used output format : C:\Program Files\FLAC\flac.exe (User Defined Encoder)
192 kBit/s
Additional command line options : -V -8 -T "artist=%a" -T "title=%t" -T "album=%g" -T "date=%y" -T "tracknumber=%n" -T "genre=%m" %s

Other options :
Fill up missing offset samples with silence : Yes
Delete leading and trailing silent blocks : No
Native Win32 interface for Win NT & 2000


Track 1
Filename Z:\Ripped CD's\01 - Once Upon a Time- The Battle of Life and Death.wav

Pre-gap length 0:00:02.00

Peak level 98.8 %
Track quality 99.9 %
Test CRC 85556AF4
Copy CRC 85556AF4
Copy OK




Is there a way to "test" mp3s to see if they were originally 320kbps mp3s (From CD extraction, to *.wav, to 320 kbps mp3 compression) or if they are just "fakes" mp3s (from CD extraction, to *.wav, to 128/160/(another bitrate) kbps mp3 compression, to 320 kbps mp3 compression)?

Thank you ahead of time to ALL of you for your cooperation =)

Best way would be to ask on the "Official" site that offers this download

If the sample count per track is not divisible by 588 it might be a transcode, and you should complain loudly and obnoxiously to the vendor from which you purchased this.

Also, simple duration vs. file size math should tell you if the bitrate reported by your software is sane.

can you expand on this?

588 samples is the size of a CD sector, so all wav files coming from CDs will have length = N * 588 samples.
But here we've got mp3, so it has to be LAME mp3 with "accurate length" tag, only then the length of mp3 is the same as the source wav.

Blimey this is very helpful, I never realised this, thanks for your words of wisdom. Sounds to me it was transcoded from a lower bit rate source to 320kbps.

Back in the good old early days of HA.org the only correct answer to questions like these would be the advise to re-encode from your own original CDs

I was an eejit. Nothing to say here. Go on, go on.

In days like today, the correct answer would be "Buy a lossless copy if you want to be sure to get perfect quality."

P.S.: Then scan it with aucdtect and kick the artist in the nuts if it was created from mp3-material.
P.S.2: If the artist is female, then you should kick her elsewhere.

1.) Yes, that is possible with mp3packer without decoding the original file. Run mp3packer without options on your file and see what happens and check the mean bitrate afterwards. if it is at 128kbps, 192kbps or so then you know it was a fake 320kbps.

2.) If the mp3packer brings a negative result (the mean bitrate doesn't change that much), now, there is still a second option. Decode the 320kbps file to WAV and then use lame 3.97 with the parameters "-V2 --vbr-new". If the mean bitrate is much below 170kbps then the sound info in the 320kbps file was minimal already resulting in a low bitrate for the encoding process. In this case you might assume that it is in fact a fake 320kbps.

That doesn't make sense. Firstly, there are plenty of recordings that produce low bitrates when encoded directly from CD using lame 3.97 -V2 --vbr-new. These recordings will also produce low bitrates if transcoded from 320kbps, whatever happened before the 320kbps encoding. Secondly, I don't believe that transcoding something necessarily lowers the (VBR) bitrate - I bet sometimes it can increase it.

Looking at the 320kbps mp3, decoded to .wav, in spectral view in an audio editor is more useful.

Of course you should buy the CD.

There are some real (i.e. pressed, not burnt) commercial CDs which are mastered from lossy sources.

Cheers,
David.

Actually, I didn't find that mp3packer could reduce the size for any of the file I tried it on, either the pure 320 Kbps or the transcoded 128 Kbps from it. I tried both with and without the -z flag.
The "real" test (without looking at a frequencies graph to try to identify a low-pass filter) was simply to do a -v2 (v0 gives more differences) encode in LAME. LAME's output is pretty good and let's you compare in which bitrates it encoded most. Between my 2 files, I see a twofold difference above 192 Kbps. The differences in file size are not as big as I'd expect:

Original (311 Kbps -> mp3packer dit it again ): 11,9 MB
Placebo (320 upcoded from 128 Kbps): 12,1 MB

v2 from original: 6,50 MB (171 Kbps)
v2 from placebo: 5,83 MB (153 Kbps)

v0 from original: 8,32 MB (218 Kbps)
v0 from placebo: 7,17 MB (188 Kbps)
v0 from placebo 128: 7,17 MB (188 Kbps)

No rule of thumb here...

Nah, the difference is quite large (20kbps or 30kbps). You already might get the impression that something was wrong with the original file.

If all you have is the "unknown" version, what does the bitrate tell you? You have no reference for comparison.

With V2 and V0, most of the bitrate "bloat" will be due to content above 16kHz and the sfb21 issue, which will be mostly removed by 128kbps encoding.

In other words, I suspect this could just be a very long winded (and nto very reliable) way of detecting a low pass filter which would be perfectly visible in spectral view.

Cheers,
David.

That's a LOT of technical stuff xDDD
Thx very much ppl, you really know a lot 'bout your topic (Audio Technology)

But, according to this info, what do you think??



This info is in a *.txt that came with the discography that I've downloaded.

So, what do you think?
Fake? Transcode?

I'm asking this because it says:



But when I play the song, it says 320 kbps :S...

Besides, in the ID3v2 Tag it says "Encoded: FairStars Audio Converter"
And I think FairStars Audio Converter can be used to transcode from a lower mp3 bitrate to a higher one...

Thank You ahead of time for your help people

If a user defined encoder was used when ripping, the bitrate settings (192 kBit/s) are not used.

FairStars Audio Converter can also be used to encode MP3 files from FLAC files...

Thx very much...
But... to summarize...
Fake 320 kbps?
Or real 320 kbps?

No way to tell at this point, if FairStars Audio Converter doesn't do gapless MP3 encoding.

Maybe programs such as aucdtect or Mr QuestionMan might help....

Aucdtect determines if audio was compressed with MP3 or not. It cannot determine if it has gone through lossy-lossy transcoding.

The EAC logs you (ajfoucault) posted only tell us that the CD was originally ripped to FLAC. They could have done anything to the album between when the log was generated and when they released it to whatever site/network you downloaded it from.

I don't think there is any way to determine if an MP3 has been transcoded from a lower bitrate or not. You could try dividing the number of samples by 588 as previously suggested, but that doesn't strike me as being reliable.

Just as a general rule...the majority of online music stores rip their audio from CD to a lossless audio codec such as FLAC, like in this case.

The text log you are providing is a EAC rip log that only lets us know they used a CD to encode to FLAC.

The probability is that they then transcoded the FLAC to mp3 and that the files you downloaded are true 320 kbs mp3 files.

Without actually being present at the time of the encode there is no way to know for sure though.

Good luck!
JXL

Nobody bothered to create and publish a tool that would do this, but technically it is possible, especially when the mp3 is such high bitrate as 320.

ajfoucault, if you are really desperate, cut a short (20-30 sec) clip from any of your mp3s and upload it here. We'll see what we can do.

you should check lowpass frequency in an audio editor like Cool Edit.

for 128kbps MP3 in most cases is 16kHz
for 192kbps MP3 in most cases is 18kHz

but don't take after this, because some old encoders like Blade don't use any lowpass filter.
besides Fraunhofer fastenc MP3 sounds really bad even at 192kbps

P.S.: for an original 320kbps MP3 lowpass should be at 20kHz or higher/

Thank you for the offering, man . First I have to figure out how to cut a little part of the mp3 so I can upload it!!!

Bah! I dunno how to do that, so I've uploaded the whole thing, sorry if it's too much (6 MB)

*removed by moderation - copyrighted material*

If you can check it and determine wether or not it's REAL 320 kbps it would be GREATLY aprreciated.



Sorry for my ignorance, but, where can I see the number of samples?? :S

Here is some info on one of the mp3s that I'm talking 'bout

MP3 Info from Winamp:
Size: 9382734 bytes
Header found at: 1640 bytes
Length: 235 seconds
MPEG-1 layer 3
320kbit, approx. 8977 frames
44100Hz Joint Stereo
CRCs: No, Copyrighted: No
Original: No, Emphasis: None

Thx ahead of time for your help.

OK, I downloaded it, but you'd better remove the link from your post. When moderators see it they will remove it anyway. (It's one of the forum rules that links to whole copyrighted songs are not allowed, only short fragments are allowed.)

In the future, you can use programs like mp3DirectCut to cut mp3s.

I will take a closer look at the file later today.

Cheers

Ooops!!!
I'm sorry, I wasn't aware of that rule...
I did not know that.
Sorry
Well, if anybody can check it and see if it is real 320 kbps mp3 it would be greatly appreciated.

BTW how can I lower my Warning Rate??

Do I have to be a good boy from now on and I'll go down on it's own, or do I have to do something else?

Thx ahead of time

For your peace of mind...

192 kbps was included in the EAC log file because the Compression settings had it at 192 kbps, but as also seen in the log file, the CD was ripped and encoded in FLAC. FLAC overrides the MP3 setting of 192 kbps (why would FLAC encode at 192 kbps anyway?)

Therefore, what you downloaded is still probably a 320 kbps MP3 transcoded from a lossless source (FLAC).

Enjoy!

I confirm what (Sly) said about the signal bandwidth (BW) of the sound contained in mp3 files with different bit rate. The less information used to reconstruct the sound (low bit rate) the low the BW of sound signal is. And this is an objective and reliable method to identify the sound.

I would say: use some Spectrum Analysis tool (for example Sony Sound Forge 8.0 has it) and check the spectrum bandwidth.

“for 128kbps MP3 in most cases” it is between15 kHz - 16kHz, around 15.3,
“for 192kbps MP3 in most cases” it is 18kHz – 19 kHz, around 18.7 kHz,
“for an original 320kbps MP3 (BW) should be 20kHz or higher”

So, if your mp3 @320 has not an 20 kHz BW it is fake.

The same way you can identify the radio source also. For example an FM radio broadcast has a signal spectrum BW around 16 kHz.

Alright, ajfoucault, I am back.

I've run a few tests on your file. I can say with certainty that it was not transcoded from mp3 of low bitrates, < 128...192.

(Even if somebody would have used a trick to keep all frequencies in such mp3, e.g. "lame -b 128 -k", even such transcoding would have been detected.)

Also it was not transcoded from wma of bitrates < 192 kbps.

However, my test is not sensitive enough to detect if files were transcoded from high bitrate source. e.g. 320 --> 320, it cannot tell.
Also I didn't bother to test mpc, ogg or aac, and I cannot test Real-audio.


In my opinion, the file looks like a genuine encoded to mp3 from flac or wav.
The frequency spectre looks alright.

A bigger concern would be that, apparently, the encoder settings were not optimal. It appears that file has no LAME tag, no bit-reservoir, possibly disabled lowpass, and short block usage seems rather low for this kind of music.

On top of that, the original music is, as usual, clipped and compressed. Maybe that is the problem you are hearing?

Another concern would be this 588 samples issue. Because the file lacks LAME tag, it is tricky to determine exact encoder/decoder offset. I cannot quite remember how to calculate precise delays/offset introduced by LAME. I tried a few combinations of numbers, but could not get a match to this "N*588" samples. But perhaps my algebra was wrong here. You should ask somebody who knows exactly these numbers.

So, this last issue still leaves a possibility that the file was transcoded from lossy source. But then it must have been a very high bitrate source. Or maybe there was no transcoding at all. Personally I wouldn't worry about it.

By the way, you can determine the number of samples in mp3 using foobar. Or decode mp3 to wav and look how many samples the wav has. (I noticed that foobar sometimes reads number of samples wrong when there is neither lame tag nor vbr tag in the file).

Cheers

Thx so much everyone!!!
Your advices and posts were precise and accurate!!!
I can rest in peace now listening to my mp3s xD

Thx everybody!!!

Out of pure curiosity, I borrowed the original CD from a friend. Sad enough EAC did not read the disc without errors on any of my drives (NEC 3520, Pioneer DVR-A07XLA, LG GSA-H12L) due to the heavy copy protection.