Advanced Audio Coding is designed to be the successor to MPEG-1 Audio Layer 3, known as MP3 format, which was specified by ISO / IEC at 11172-3 (MPEG-1 Audio) and 13818-3 (MPEG-2 Audio).
Blind tests in the late 1990s showed that AAC demonstrated higher sound quality and transparency than MP3 for files encoded with the same bitrate.
The improvements include:
higher sampling frequencies (8-96 kHz) than MP3 format (16 to 48 kHz);
up to 48 channels (MP3 supports up to two channels in MPEG-1 mode and up to 5.1 channels in MPEG-2 mode);
Arbitrary bit rates and variable frame length. Standardized constant bit rate with bit deposit);
higher efficiency and simpler filter bank (instead of hybrid MP3 encoding, AAC uses pure MDCT);
higher coding efficiency for stationary signals (AAC uses a block size of 1024 or 960 samples, allowing more efficient coding of sample blocks than MP3 576);
higher coding precision for transient signals (AAC uses a block equal to 128 or 120 samples, allowing more precise coding of blocks of MP3 192 samples);
possibility of using derivatives of the Kaiser-Bessel window function to eliminate spectral dispersion at the expense of enlarging the main lobe;
much better management of audio frequencies above 16 kHz;
more flexible joint stereo (different methods can be used in different frequency ranges);
additional modules (tools) added to increase compression efficiency: TNS, Back Prediction, PNS, etc. These modules can be combined to form different encoding profiles.
In general, the AAC format allows developers more flexibility in codec design than MP3 and corrects many of the design choices made in the original MPEG-1 audio specification. This increased flexibility often leads to multiple simultaneous encoding strategies and consequently more efficient compression. However, in terms of whether AAC is better than MP3, the advantages of AAC are not entirely conclusive, and the MP3 specification, while dated, has proven surprisingly robust despite notable flaws. AAC and HE-AAC are better than MP3 at low bit rates (typically less than 128 kilobits per second). This is especially true at very low bit rates where superior stereo, pure MDCT encoding, and better transform window sizes let MP3 compete.
While the MP3 format has almost universal hardware and software support, mainly because MP3 was the format of choice during the crucial early years of music sharing / distribution over the Internet, AAC is a strong competitor due to some unwavering support from the industry.
How AAC works
AAC is a wideband audio coding algorithm that takes advantage of two main coding strategies to dramatically reduce the amount of data required to represent high-quality digital audio:
Components of the signals that are perceptually irrelevant are discarded.
Excess in the encoded audio signal is removed.
The actual encoding process consists of the following steps:
The signal is converted from the time domain to the frequency domain using the Forward Modified Discrete Cosine Transform (MDCT). This is done using filter banks that take an adequate number of time samples and convert them to frequency samples.
The signal in the frequency domain is quantized based on a psychoacoustic model and encoded.
Internal error correction codes are added.
The signal is stored or transmitted.
To avoid corrupted samples, a modern implementation of the luhn mod N formula is applied to each frame.
The MPEG-4 audio standard does not define a single or small set of highly efficient compression schemes, but rather a complex set of tools to perform a wide range of bitrate encoding operations, from low speech to audio encoding. high quality and musical synthesis.
The ‘MPEG-4 family audio coding algorithm covers the range from low speech coding bit rate (up to 2 kbit / s) to high quality audio coding (at 64 kbit / s per channel and higher).
AAC offers sample rates between 8 kHz and 96 kHz and any number of channels between 1 and 48.
In contrast to MP3’s hybrid filter bank, AAC uses Modified Discrete Cosine Transform (MDCT) in conjunction with increasing window lengths of 1024 or 960 points.
