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: MP3 Duration Problem in Win7/XP (Read 4571 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

MP3 Duration Problem in Win7/XP

Hi Folks,

First ever post, so please bear with me 

I'm not an audio expert (background is electronics) but I've been doing some DIY forensics on a supposed 'contract call' MP3 recording for an issue my wife had last year, we know that the recording has been heavily edited. We're quite safe from the scam now, but I wanted to prove that my wife's version of events was correct for her sake. I've managed to pull the thing apart on a number of angles,
for example, ENF extraction that a forensics lab based a couple of miles away from the offending rogue business, said was non-existent, multiple ENF's, phase errors in the ENF, spectrogram edits, multiple transcoding upscaling, noise floor problems, power harmonics, filtering etc and the evidence is quite overwhelming that
the file has been tampered with. The call recording duration is about 8m 45s long according to most tools that I have access to. The actual original total call length according to the telephone log, was about 38 minutes long.
That wouldn't be a problem as these people only record the actual 'contract' part of the call, however, when doing some tests with the recording on one of our old Lab WinXP PC's at work, the system properties report the mp3 duration is 38m 24s and shows that same duration in WMP etc, not 8m 45s, but only plays for 8m 45s.
I've tried using every meta data analyser that I can get hold of to try and determine where the 38m 24s is being stored or originates from, checked the mp3 size calculation on the HA site, which tallies with Win7 etc as approximately 8m45s.
So there's the question, and this is where I hope you folks may be able to shed some light as to whether it's pure coincidence that the reported duration in WinXP is about the same as the actual original unedited overall call length and it's just a quirk of WinXP, is WinXP finding some hidden metadata in the 'unknown stream' (see report below) or wherever or does WinXP use a method other than the metadata that our Win7/8 based systems use to calculate duration?
I've checked the file with MediaInfo, MP3 Diags, Adobe Audition V1.5, MP3Utility,Winamp etc and they show some oddities in their results:

MediaInfo:
      General
      Count :308
      Count of stream of this kind :1
      Kind of stream :General
      Kind of stream :General
      Stream identifier :0
      Inform :Complete name : K:\Audio Tests\New folder\Copy of Originals\TBS - CC ELEC.mp3 Format : MPEG Audio File size : 8.02 MiB
      Duration : 8mn 45s Overall bit rate mode : Constant Overall bit rate : 128
      Kbps Genre : Abstract Writing library : LAME3.82
      Count of audio streams :1
      Audio_Format_List :MPEG Audio
      Audio_Format_WithHint_List :MPEG Audio
      Audio codecs :MPEG-2 Audio layer 3
      Complete name :K:\Audio Tests\New folder\Copy of Originals\TBS - CC ELEC.mp3
      Folder name :K:\Audio Tests\New folder\Copy of Originals
      File name : TBS - CC ELEC
      File extension :mp3
      Format :MPEG Audio
      Format :MPEG Audio
      Format/Extensions usually used :m1a mpa1 mp1 m2a mpa2 mp2 mp3
      Commercial name :MPEG Audio
      Internet media type :audio/mpeg
      Codec :MPEG Audio
      Codec :MPEG Audio
      Codec/Extensions usually used :m1a mpa1 mp1 m2a mpa2 mp2 mp3
      File size :8409098
      File size :8.02 MiB
      File size :8 MiB
      File size :8.0 MiB
      File size :8.02 MiB
      File size :8.020 MiB
      Duration :525557
      Duration :8mn 45s
      Duration :8mn 45s 557ms
      Duration :8mn 45s
      Duration :00:08:45.557
      Duration :00:08:45.557
      Overall bit rate mode :CBR
      Overall bit rate mode :Constant
      Overall bit rate :128000
      Overall bit rate :128 Kbps
      Stream size :177
      Stream size :177 Bytes (0%)
      Stream size :177 Bytes
      Stream size :177 Bytes
      Stream size :177 Bytes
      Stream size :177.0 Bytes
      Stream size :177 Bytes (0%)
      Proportion of this stream :0.00002
      Genre :Abstract
      File creation date :UTC 2015-12-04 14:49:12.484
      File creation date (local) :2015-12-04 14:49:12.484
      File last modification date :UTC 2015-03-03 13:50:00.000
      File last modification date (local) :2015-03-03 13:50:00.000
      Writing library :LAME3.82
      Writing library :LAME3.82

      Audio
      Count :254
      Count of stream of this kind :1
      Kind of stream :Audio
      Kind of stream :Audio
      Stream identifier :0
      Inform :Format : MPEG Audio Format version : Version 2 Format profile :
      Layer 3 Duration : 8mn 46s Bit rate mode : Constant Bit rate : 128 Kbps
      Channel(s) : 2 channels Sampling rate : 22.05 KHz Compression mode : Lossy
      Stream size : 8.02 MiB (100%) Writing library : LAME3.82
      Format :MPEG Audio
      Commercial name :MPEG Audio
      Format version :Version 2
      Format profile :Layer 3
      Internet media type :audio/mpeg
      Codec :MPA2L3
      Codec :MPEG-2 Audio layer 3
      Duration :526759
      Duration :8mn 46s
      Duration :8mn 46s 759ms
      Duration :8mn 46s
      Duration :00:08:46.759
      Duration :00:08:46.759
      Bit rate mode :CBR
      Bit rate mode :Constant
      Bit rate :128000
      Bit rate :128 Kbps
      Channel(s) :2
      Channel(s) :2 channels
      Sampling rate :22050
      Sampling rate :22.05 KHz
      Samples count :11615040
      Frame count :20165
      Compression mode :Lossy
      Compression mode :Lossy
      Stream size :8408921
      Stream size :8.02 MiB (100%)
      Stream size :8 MiB
      Stream size :8.0 MiB
      Stream size :8.02 MiB
      Stream size :8.019 MiB
      Stream size :8.02 MiB (100%)
      Proportion of this stream :0.99998
      Writing library :LAME3.82
      Writing library :LAME3.82

MP3 Diags Report:
Address 0x0: 49 Bytes (D) ox31(H) Stream Details: ID3V2.3.0 Padding=0, unsynch=no, frames:POPM, TLEN="2304256"
Address 0x31: 8408921 Bytes (D) 0x804f59(H) Stream Details: 8.45, MPEG-2 layer III, stereo, 22050Hz, 128000bps CBR, CRC=no, frame count=20119; last frame located at 0x804de8
ID3V2 tag doesn't have an APIC frame (used to store images).
Invalid ID3V1 tag. Invalid characters in year field.
Unknown stream found size: 128 Bytes (D) 0x80 (H); begins with "TAG.........(58 zeroes)..." (54 41 47 00 00...(58 zeroes total)...00 00)

Adobe Audition V1.5 Report:
Filename:   TBS - CC ELEC.mp3
Folder:   K:\Audio Tests\New folder\Copy of Originals
File Type:   22050Hz, 16-bit, Stereo
Uncompressed Size:   44.2 MB (46,354,448 bytes)
File Format:   mp3PRO® (FhG)
   MPEG Layer-3, 128 Kbps (Stereo)
Size on Disk:   8.01 MB (8,409,098.00 bytes)
Last Written (local):   3/3/2015  13:50:00.000
Length:   8:45.560
   11,588,612 samples

MP3Utility:
Summary: 20,119 total frames processed (19,298 padded, 821 unpadded).  Bitrate is constant

Winamp:
Size: 8408970 bytes
Header found at: 49 bytes
Length: 526 seconds
MPEG-2 layer 3
128kbit, approx. 20117 frames
22050Hz Stereo
CRCs: No, Copyrighted: No
Original: Yes, Emphasis: None

So here's the  thing, has anyone got any ideas as to where that 38m 24s is coming from?

Many thanks for any help or ideas,

Mike

MP3 Duration Problem in Win7/XP

Reply #1
Windows has a lot of trouble estimating the length of MP3 files.  Its often widely inaccurate.

MP3 Duration Problem in Win7/XP

Reply #2
Address 0x0: 49 Bytes (D) ox31(H) Stream Details: ID3V2.3.0 Padding=0, unsynch=no, frames:POPM, TLEN="2304256"

Found it.

ID3v2 TLEN represents the length in milliseconds. The value 2304256 represents 38 minutes and 24 seconds.


MP3 Duration Problem in Win7/XP

Reply #4
We posted at very nearly the same time and you haven't replied to me yet, so I'm guessing you didn't see my post...

Address 0x0: 49 Bytes (D) ox31(H) Stream Details: ID3V2.3.0 Padding=0, unsynch=no, frames:POPM, TLEN="2304256"

The value 2304256 represents 38 minutes and 24 seconds.


MP3 Duration Problem in Win7/XP

Reply #5
The simplest solution would be to convert to WAV.    It doesn't seem like there's any reason you need MP3.

Since you have Adobe Audition, you can use it to make a WAV file.  Or, you can make new MP3 from Audition.  (Re-encoding to MP3 is an additional generation of lossy compression, but it should be OK for a voice/phone recording.)

I've seen that problem with variable bitrate files, but you've got constant bitrate.  There is a tool called VBRfix, but I don't know if it will do anything with your CBR file, and I thought  MP3diags could make the same fix.

MP3 Duration Problem in Win7/XP

Reply #6
It looks like VBRfix is geared towards fixing the VBR header, but the 38-minute length is coming from an ID3 tag. A tag editor should be able to locate the offending data. (My favorite is MP3tag, but I've never used it to modify a TLEN tag so I have no idea how it handles that.)

MP3 Duration Problem in Win7/XP

Reply #7
We posted at very nearly the same time and you haven't replied to me yet, so I'm guessing you didn't see my post...

Address 0x0: 49 Bytes (D) ox31(H) Stream Details: ID3V2.3.0 Padding=0, unsynch=no, frames:POPM, TLEN="2304256"

The value 2304256 represents 38 minutes and 24 seconds.



Hi Octocontrabass, apologies, the evening got hijacked last night just after I posted so I've only just caught up. Thanks for the super quick reply (and that goes to everyone).

Ahhh! Of course, I missed that! I assumed, without double checking and actually doing the maths, that the TLEN was being converted to the 8m 45s and it is of course as you say 38m 24s, thanks for pointing me back in the right direction 

Re. actually changing the value, that's not quite the direction I'm going in with this, though thanks to both DVDdoug and Octocontrabass for suggesting using MP3diags to correct the 'problem'.
What I'm actually doing is ripping apart the file for evidence of editing (actually two files, one is a 'contract call' recording (the MP3) and one is a 'comfort call' recording (WAV)) so that I can reply to our Trading Standards Officer dealing with the case with a technical response to the Forensic lab' that processed the original recordings supplied to us.
The issue is that, as I say, we know that the recordings are falsified, and they were sent in good faith to the lab' for them to prove this, but it has become quite obvious from the absurd dismissal that the lab' has, shall we say, 'connections', to the energy company concerned.
The observant (and I know that includes you folks) will notice the last mod' date. It took a number of months to get the recordings and get us clear from this company and I wasn't going to pursue it any further as it was of no further consequence to us and the company owners are very powerful ex-convicts who employ a lot of people and have had high ranking UK Government Ministers visit them.
That doesn't alter the facts though that they are part of a major investigation for company impersonation fraud, callerID spoofing, rogue contracts and so forth, literally putting people out of business and on the street with their business model.
The initial analysis of the recordings was a breakthrough in the case for TS, who have been working this for a number of years now, as most small businesses being targeted don't have people with a technical background lurking in the shadows. The case stagnated when my evidence was discounted by the lab' and I only decided to get involved again after hearing of further victims to this very clever fraud that exploits a UK loophole in contract law.
So what I have done, as the TS officers don't understand the technical aspects of the files, and as I say, I'm not an audio expert but I do have >30 years background in electronics, is research forensic techniques for edit and tampering detection used by the pro's (Dr. Catalin Grigoras, Christopher William Jenkins, Marcin Micha?ek etc) and analyse the recordings to produce a counter report that is going to be analysed by a third party University Forensics Faculty near where I live. This will give TS the opportunity to get verification that the original lab' has 'issues' and hopefully get the case going again before too many more businesses fall prey to these vultures.
To be honest, it would be a whole lot easier if I was an audio expert, but it has been very interesting to get involved with this and get the opportunity to open up a right can of worms.

Thanks again for your input as it has enabled me to confirm that the original call length was as we thought and had put in our statements, and that the recording supplied is a sub-part of that original recording. This means, I am led to believe, that what should have been a legally binding contract, would not be admissible in a Court however as it is yet further evidence of editing.

 

Re: MP3 Duration Problem in Win7/XP

Reply #8
why don't you submit this as evidence to some kind of police force's audio forensics?  i don't understand why you want to do this all yourself without the proper training?  But I'm not against the DIY philosophy, don't get me wrong.  It just seems that if this is for a court case or prosecuting criminals, you need to bring in somebody who is an expert so no mistakes legal or technical are made. 

It's just my 2 cents.  Good luck either way. 
Be a false negative of yourself!