Low-delay transform coding using the MPEG-H 3D audio codec

Recently, the ISO/IEC MPEG-H 3D Audio standard for perceptual coding of one or more audio channels has been finalized. It is a little-known fact that, particularly for communication applications, the 3D Audio core-codec can be operated in a low-latency configuration in order to reduce the algorithmic coding/decoding delay to 44, 33, 24, or 18 ms at a sampling rate of 48 kHz. This paper introduces the essential coding tools required for high-quality low-delay coding - transform splitting, intelligent gap filling, and stereo filling - and demonstrates by means of blind listening tests that the achievable subjective performance compares favorably with e. g. that of HE-AAC even at low bit-rates.