Diego, tell us a bit about yourself. What do you do, and what does a typical day in your life look like?
Currently, I am a professor in the Telecommunications Engineering Faculty, at Universidad Militar Nueva Granada (Bogotá, Colombia). I teach about mobile communications and digital signal processing.My latest projects have been about image and audio watermarking; my current project is about audio forensics. In a typical day, I have classes and I do research.
You've recently released a paper about the authenticity verification of audio signals based on fragile watermarking for audio forensics. For those who don't know, could you briefly describe fragile watermarking?
Generally, watermarking is a group of techniques focused on adding some content (mark) to a host file. For instance, think of a photographer who adds his name (mark) to each photo (host) as a copyright signature. If the photographer wants the mark to persist after modifications in the photography, he should choose robust watermarking. But, if the purpose is to identify if the marked file has been tampered and where, he should choose fragile watermarking. The first one is typically applied for copyright protection. The second one is useful for tampering detection.
What prompted you to choose this as a research subject?
Choosing the research subject, we realized that in our country there are a lot of audio recordings given as evidence, that finally are not used in the court, because they are dismissed. For instance, these kinds of proofs are given in cases such as kidnapping, extortion or intimate partner violence. In this context, we wonder what we could bring from the point of view of signal processing, to offer greater reliability to the evidence.
Can you outline for us the aims and results of your research?
We have researched on audio integrity for the last two years, focused on two main points of view: perceptual hash functions and fragile watermarking.
In the first case, we have developed a hash function for audio forensics which is non-sensitive to perceptual modifications like quantization, but highly sensitive to other kinds of manipulations like cropping.
In the second case, we have proposed a scheme of fragile watermarking that allows people to detect if the audio has been tampered with, as well as where the signal has changed. Both solutions can be used to support chain of custody.
You can access the full article through this link. It is free to access until October 31, 2017.
What are some of the most pertinent challenges in the field of audio forensics at the moment, and how might these be addressed?
In my opinion, the most important challenge in the field of audio forensics lies in blind analysis of the authenticity of the audio proof. What does it mean? Without knowledge of the original audio, the researcher must identify if some type of manipulation has been applied. This aspect is very important because there are a lot of tools for computers, mobile phones and tablets that allow someone to tamper with an audio file without providing a perceptible trace of the manipulation.
In what circumstances might a digital forensic examiner come across audio evidence that has been tampered with?
There are two important scenarios. The first one is related to the recording process and the second one to the chain of custody.
In the first scenario the victim deliberately manipulates the audio recording before delivering it as evidence. In the second one, once the evidence has been delivered to the legal authority, a trespasser can access the storage system and replace the original file with a tampered version. Regardless of the above, the legal authority must be sure whether the audio is unaltered or tampered.
Are you planning any new research at the moment?
Currently, we are working on methods to generate hash codes of the audio signals in the same moment that the audio is recorded. Specifically, we are working on an app for audio recording with a seal of authenticity.
Finally, when you're not researching, what do you enjoy doing in your spare time?
I usually share time with my family. Also, I enjoy watching TV series, reading literature books and playing basketball.
Diego received his M.Sc. degree in Telecommunications Engineering from National University of Colombia in 2010, and a Ph.D. in Advanced Computing from the Technical University of Madrid in 2015. In a broad sense, his research interests include signal processing and remote sensing. Diego's latest paper is free to access online until October 31, 2017. You can find out more about Diego and follow his work on ResearchGate.