INTERPOL survey of the use of speaker identification by law enforcement agencies

Morrison, Geoffrey Stewart; Sahito, Farhan; Jardine, Gaëlle; Djokic, Djordje; Clavet, Sophie; Berghs, Sabine; Dorny, Caroline Goemans

doi:10.1016/j.forsciint.2016.03.044

Cited by 51 publications

(29 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the Mel domain, the filters are equally spaced in a frequency band [29]. After signal filtering, the DCT (discrete cosine transform) is applied to the logarithm of the above data [30], in order to de-correlate them. Finally, a lifter is applied to obtain the MFCC coefficients.…”

Section: Mfccmentioning

confidence: 99%

Robust Speech Hashing for Digital Audio Forensics

2019

View full text Add to dashboard Cite

The verification of the integrity and authenticity of multimedia content is an essential task in the forensic field, in order to make digital evidence admissible. The main objective is to establish whether the multimedia content has been manipulated with significant changes to its content, such as the removal of noise (e.g., a gunshot) that could clarify the facts of a crime. In this project we propose a method to generate a summary value for audio recordings, known as hash. Our method is robust, which means that if the audio has been modified slightly (without changing its significant content) with perceptual manipulations such as MPEG-4 AAC, the hash value of the new audio is very similar to that of the original audio; on the contrary, if the audio is altered and its content changes, for example with a low pass filter, the new hash value moves away from the original value. The method starts with the application of MFCC (Mel-frequency cepstrum coefficients) and the reduction of dimensions through the analysis of main components (principal component analysis, PCA). The reduced data is encrypted using as inputs two values from a particular binarization system using Collatz conjecture as the basis. Finally, a robust 96-bit code is obtained, which varies little when perceptual modifications are made to the signal such as compression or amplitude modification. According to experimental tests, the BER (bit error rate) between the hash value of the original audio recording and the manipulated audio recording is low for perceptual manipulations, i.e., 0% for FLAC and re-quantization, 1% in average for volume (−6 dB gain), less than 5% in average for MPEG-4 and resampling (using the FIR anti-aliasing filter); but more than 25% for non-perceptual manipulations such as low pass filtering (3 kHz, fifth order), additive noise, cutting and copy-move.

show abstract

Section: Mfccmentioning

confidence: 99%

Robust Speech Hashing for Digital Audio Forensics

2019

View full text Add to dashboard Cite

show abstract

“…While both DNNs and GMMs aim at incorporating phonetic information of the phrase with these posteriors, model-based SID approaches ignore the sequence information of the phonetic units of the phrase. SIIP overcomes this problem by applying a dynamic time warping architecture using speaker-informative features [9]. Further, also a combination of SID with other modalities such as with automatic speech recognition or keyword 1 http://www.siip.eu spotting engine allows the use of content information in speaker identification.…”

Section: A Speaker Identificationmentioning

confidence: 99%

“…Subsequent telephone interviews were held with 40 survey respondents. A paper was also submitted and published in the special issue of "Forensic Science International" [9].…”

Section: A Siip Survey Questionnairesmentioning

confidence: 99%

Towards a Breakthrough Speaker Identification Approach for Law Enforcement Agencies: SIIP

Khelif

Mombrun

Backfried³

et al. 2017

2017 European Intelligence and Security Informatics Conference (EISIC)

View full text Add to dashboard Cite

Abstract-Thispaper describes SIIP (Speaker Identification Integrated Project) a high performance innovative and sustainable Speaker Identification (SID) solution, running over large voice samples database. The solution is based on development, integration and fusion of a series of speech analytic algorithms which includes speaker model recognition, gender identification, age identification, language and accent identification, keyword and taxonomy spotting. A full integrated system is proposed ensuring multisource data management, advanced voice analysis, information sharing and efficient and consistent man-machine interactions.

show abstract

“…These experts are often phoneticians who have obtained additional training in forensic science, or engineers or computer scientists who have obtained additional training in phonetics and forensic science [14]. World-wide surveys on forensic speech investigation [15,16] show that the most popular approach employed in most countries are auditory analysis, together with acoustic-phonetic analysis. Reich [4] conducted an experiment where two groups of listeners, naive (undergraduate students) and sophisticated (doctoral students and professors in speech and hearing sciences), discriminated disguised and undisguised speech created by forty male speakers.…”

Section: Introductionmentioning

confidence: 99%