Intelligent gap filling in perceptual transform coding of audio

Intelligent Gap Filling (IGF) denotes a semi-parametric coding technique within modern codecs like MPEG-H 3D Audio or the 3GPP EVS codec. IGF can be applied to fill spectral holes introduced by the quantization process in the encoder due to low-bitrate constraints. Typically, if the limited bit budget does not allow for transparent coding, spectral holes emerge in the high-frequency (HF) region of the signal first and increasingly affect the entire upper spectral range for lowest bitrates. At the decoder side, such spectral holes are substituted via IGF using synthetic HF content generated in a semi-parametric fashion out of low-frequency (LF) content, and post-processing controlled by additional parametric side information. This paper provides an overview of the principles and functionalities of IGF and presents listening test data assessing the perceptual quality of IGF coded audio material.