Notifications
Clear all

Audioforensic

6 Posts
4 Users
0 Likes
530 Views
 dega
(@dega)
Posts: 261
Reputable Member
Topic starter
 

dear all, I am studying a technical review. I would like to know which are the tecniques use to analyze a speech and verify it isn't a cut and pasta.
thanks

 
Posted : 15/09/2017 2:19 pm
gungora
(@gungora)
Posts: 33
Eminent Member
 

SWGDE's best practices document for digital audio authentication can be a good starting point

https://www.swgde.org/documents/Current%20Documents/SWGDE%20Best%20Practices%20for%20Digital%20Audio%20Authentication

 
Posted : 15/09/2017 5:58 pm
jpickens
(@jpickens)
Posts: 130
Estimable Member
 

dear all, I am studying a technical review. I would like to know which are the tecniques use to analyze a speech and verify it isn't a cut and pasta.
thanks

if it's digital audio, I'm pretty sure it will not do well in a ragu or bolognese sauce.

 
Posted : 15/09/2017 6:11 pm
 dega
(@dega)
Posts: 261
Reputable Member
Topic starter
 

SWGDE's best practices document for digital audio authentication can be a good starting point

https://www.swgde.org/documents/Current%20Documents/SWGDE%20Best%20Practices%20for%20Digital%20Audio%20Authentication

Great! It's whats I need! Perfect thanks

 
Posted : 16/09/2017 2:15 pm
(@verdenforensics)
Posts: 1
New Member
 

There are many, many different methods available to determine the authenticity of an audio file. I would begin with some file structure analysis (determine that the sample-rate, bit-depth, number of channels) is consistent with the file type. Then there is the header, which may contain traces of software such as Abode Audition if the file has been opened and saved out of such software, indicating it may have been tampered with.

Follow this up with some analysis of the waveform (including FFT, phase consistency) . You can then look into areas such as compression analysis (to determine if the audio has been re-compressed, indicating it has been saved at least twice). DC Offset (available to view in Adobe Audition), b**t Splice detection, Long Term Average Spectrum (LTAS), Sorted Spectrum), ENF analysis are a few other methods you may want to look into.

I have an article regarding audio forensics in 2017 in the latest edition of Digital Forensics Magazine which goes into these areas in more detail. Let me know if you would like me to send over the section which deals with audio authentication. And as a previous poster stated, SWIGDE and ENFSI do have some great guidelines on the subject.

 
Posted : 18/09/2017 8:16 am
 dega
(@dega)
Posts: 261
Reputable Member
Topic starter
 

verdenforensic you've got a private mail

 
Posted : 26/09/2017 8:16 am
Share: