Audio Lossless Coding

MPEG-4 Audio Lossless Coding (MPEG-4 ALS) is an extension to the MPEG-4 Part 3 audio standard designed to enable lossless data compression of digital audio. Developed to retain original audio samples with complete accuracy, the standard offers a high-performance alternative to perceptual audio coders that discard information. Its specification was finalised in 2005 and incorporated into international audio standards shortly afterwards, forming part of a broader set of MPEG-4 audio technologies used in professional and consumer environments.
MPEG-4 ALS was further consolidated and republished within the revised MPEG-4 Audio standard in 2019, reinforcing its relevance as a method for mathematically reversible audio compression. Although technically sophisticated and capable of high compression efficiency, its adoption has remained limited compared with more widespread lossless formats.

Technical structure and predictive coding model

MPEG-4 ALS is built on the principle of predictive audio coding, using both short-term and long-term prediction stages to remove redundancy from audio waveforms.
The short-term predictor resembles the linear predictive coding methods employed in certain other lossless formats. It uses a quantised linear predictive coding filter to model the local structure of the signal, with the residual error coded using Golomb coding or Block Gilbert–Moore Coding. This stage is effective for capturing correlations between adjacent samples and reducing the magnitude of the residual to be stored.
The long-term predictor models periodicities and harmonic relationships within the signal. It operates by generating five long-term weighted residues, each associated with an independent lag value, often ranging into the hundreds of samples. This structure is particularly suitable for signals with strong harmonic content, such as musical instruments or the human voice, where multiples of a fundamental frequency appear in phase-coherent form.
By combining these predictive stages, MPEG-4 ALS achieves efficient lossless compression while preserving exact waveform reconstruction at decoding.

Core features and capabilities

The standard incorporates a range of advanced features that support professional audio work and flexible deployment scenarios:

• Support for high-resolution audio: pulse-code modulation signals up to 32-bit depth, including fixed-point and floating-point formats.
• Broad sampling-rate compatibility: allowing the encoding of audio across conventional and specialised frequency ranges.
• Extensive multichannel support: accommodating up to 65,536 channels and therefore suitable for multitrack recording environments and spatial audio research.
• Random access: enabling rapid seeking within encoded material, a property important for editing and media applications.
• MP4 container integration: allowing ALS streams to be multiplexed with video, timed metadata and other audio components within the MP4 file format.
• Defined audio profiles: including the ALS Simple Profile for constrained implementations where reduced complexity is required.

These capabilities make the format adaptable to varied use cases, ranging from archival preservation to complex production workflows.

Software implementation and ecosystem

Despite its technical strengths, MPEG-4 ALS has seen relatively modest adoption, a situation often attributed to the limited availability of encoders and decoders. A reference implementation of both encoding and decoding functionality is provided through the official mp4als software package, which has appeared in successive international standard amendments.
Support within third-party applications has been sporadic. A decoder plug-in has been available for the Winamp media player, and in 2009 an ALS decoder was incorporated into the development branch of the FFmpeg open-source project, although only partial format support was implemented. Broader uptake has been constrained by competition from more widely supported lossless codecs and by the specific computational demands associated with advanced predictive coding.

Development history and standardisation process

The origins of MPEG-4 ALS date to 2002, when the Moving Picture Experts Group issued a call for proposals for a lossless audio coding standard. Seven candidate systems were submitted and evaluated on the basis of compression ratio, computational complexity and flexibility.
By mid-2003, Lossless Predictive Audio Compression (LPAC), a system developed with significant academic and industrial collaboration, was selected as the basis for the emerging standard. Subsequent development involved contributions from research groups and companies including Technische Universität Berlin, RealNetworks and Nippon Telegraph and Telephone. Their work shaped the reference model that would become the formalised MPEG-4 ALS specification.
The completed standard was incorporated as an amendment to MPEG-4 Part 3 in 2006, with later revisions consolidating the format within the 2019 edition of the MPEG-4 Audio specification.