Methods for identifying traces of compression in audio
Samet Hiçsönmez, Erkam Uzun, Hüsrev Taha Sencar
Abstract
Samet Hiçsönmez, Erkam Uzun, Hüsrev Taha Sencar
Abstract
Compression history of an audio may reveal very useful information when traces of tampering has to be investigated or quality of an audio has to be evaluated. Motivated by this, we introduce two methods that can discriminate between single and double compressed audio and can identify compression codec and bit rate of an audio. The first method utilizes audio quality measures to realize this and operates on decoded audio. The second method, alternatively, works on coded audio, effectively the audio bit stream, and characterizes randomness and chaotic properties of the bit stream to achieve these tasks. Unlike the existing work in the literature, which are proposed mainly for MP3 encoded audio, both methods can be applied to all encoding formats. Extensive tests have been performed to test the performance of both methods under various settings. Results show that both methods can be very reliably used to obtain information on compression history of an audio.
OpenAlex reports 19 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Compression history of an audio may reveal very useful information when traces of tampering has to be investigated or quality of an audio has to be evaluated. Motivated by this, we introduce two methods that can discriminate between single and double compressed audio and can identify compression codec and bit rate of an audio. The first method utilizes audio quality measures to realize this and operates on decoded audio. The second method, alternatively, works on coded audio, effectively the audio bit stream, and characterizes randomness and chaotic properties of the bit stream to achieve these tasks. Unlike the existing work in the literature, which are proposed mainly for MP3 encoded audio, both methods can be applied to all encoding formats. Extensive tests have been performed to test the performance of both methods under various settings. Results show that both methods can be very reliably used to obtain information on compression history of an audio.
Key concepts: Computer science, Sound quality, Codec, Speech coding, Dynamic range compression, Data compression, Speech recognition, Encoding (memory)