Meridian Lossless Packing

Meridian Lossless Packing (MLP) is a proprietary lossless audio compression algorithm developed by the British audio company Meridian Audio, designed to encode high-resolution pulse-code modulation (PCM) audio data with bit-for-bit accuracy upon decoding.^[1] Introduced in 1997, MLP enables efficient storage and transmission of uncompressed-quality audio by reducing data redundancy through techniques such as waveform prediction, entropy coding, and matrixing, typically achieving compression ratios of 50-55% for formats like 96 kHz/24-bit multichannel audio.^[2]^[3] MLP was specifically optimized for emerging digital media standards, serving as the mandatory encoding method for DVD-Audio discs to fit high-fidelity, up to 24-bit/192 kHz, multichannel content within bitrate constraints, such as extending playback time to 74 minutes for six channels on a single-layer DVD.^[1] Its robust design includes built-in error detection and recovery, allowing restarts within 10-30 milliseconds and cascadability for multiple encode-decode cycles without cumulative degradation, making it suitable for professional studio workflows and consumer playback.^[1] In 1997, Meridian licensed the technology to Dolby Laboratories, where it forms the core of Dolby TrueHD, a lossless codec supporting up to 32 full-range channels and higher bitrates for Blu-ray Disc and HD DVD, enhancing home theater audio fidelity.^[2]^[3] Technically, MLP supports sampling rates from 44.1 kHz to 192 kHz, up to 64 channels, and 24-bit precision, with decoder implementations requiring modest processing power—such as handling six channels at 96 kHz/24-bit on a single digital signal processor (DSP).^[1] The bitstream incorporates metadata for cueing, dynamic range control, and additional data channels, while variable or fixed data rates help manage peak demands, capping at around 9.6 Mbps for DVD-Audio applications.^[3] Beyond optical media, MLP has influenced archival audio formats and remains relevant in high-end audio systems, underscoring Meridian's contributions to digital sound preservation and delivery.^[2]

History and Development

Origins and Creation

Meridian Audio Limited, a British audio technology company based in Cambridge, England, initiated the development of Meridian Lossless Packing (MLP) in the mid-1990s as a proprietary audio coding scheme designed to address the challenges of storing high-fidelity digital audio on emerging optical media formats.^[1] The project spanned approximately three years, culminating in the release of provisional technical data in June 1998, which outlined the system's architecture and capabilities.^[1] The primary motivation for creating MLP was to enable lossless compression of multi-channel pulse-code modulation (PCM) audio at high sample rates and bit depths, such as 24-bit/96 kHz, allowing this data to fit efficiently on storage-limited discs like DVDs while preserving bit-for-bit audio quality without any perceptual loss.^[1] This innovation targeted the growing demand for surround sound and high-resolution formats in consumer audio, where traditional uncompressed PCM would exceed practical storage capacities for multi-channel content.^[1] Initially named Packed PCM (PPCM), the technology remained proprietary to Meridian during its early phases, with the company integrating prototype decoders into its surround sound processors starting in June 1998.^[1] By 1999, MLP had been selected as the standard lossless encoding method for DVD-Audio, with further refinements documented in technical releases around 2000 to support its integration into that format.^[4] This foundational work later extended to high-definition media, where MLP formed the basis for Dolby TrueHD in Blu-ray audio.^[2]

Adoption in Audio Standards

In 1998, Meridian Audio licensed its Meridian Lossless Packing (MLP) technology to Dolby Laboratories, enabling broader adoption in consumer audio formats. This agreement positioned Dolby as the primary licensor for MLP implementations, facilitating its integration into high-resolution audio standards. The technology was specifically tailored for efficient, lossless compression of multichannel PCM audio, supporting up to 24-bit/192 kHz signals without data alteration.^[5] The DVD Forum, an international standards body comprising electronics manufacturers and content providers, incorporated MLP into the DVD-Audio specification, which was approved on February 8, 1999. Under this standard, MLP became the mandatory decoding format for all DVD-Audio players to handle high-resolution, lossless audio tracks, ensuring compatibility for up to 5.1 channels at rates exceeding those of standard DVD-Video. While LPCM remained an option for basic stereo content, MLP's inclusion was pivotal for advanced multichannel and high-sample-rate applications, with the first commercial DVD-Audio discs released by major labels like Warner Music Group in November 2000.^[6] Building on this foundation, MLP evolved into Dolby TrueHD, a core component of the Blu-ray Disc format developed by the Blu-ray Disc Association (BDA). Announced in 2004 as part of the high-definition optical disc standards, Dolby TrueHD extended MLP's capabilities to support up to 14 channels at 192 kHz/24-bit, with Blu-ray players required to decode it for lossless playback. The format launched commercially with Blu-ray discs in 2006, marking MLP's transition to next-generation media and solidifying its role in professional and consumer high-fidelity audio delivery.^[7] Meridian's core patents underpinning MLP, such as GB2323754 for lossless compression using IIR prediction filters filed in 1997, were licensed through Dolby and expired around 2017, allowing for unrestricted implementations post-patent term. This intellectual property framework, combined with standards body endorsements, propelled MLP from a proprietary innovation to an industry cornerstone for bit-exact audio preservation.^[8]

Technical Overview

Core Encoding Process

Meridian Lossless Packing (MLP), developed by Meridian Audio in the late 1990s, processes uncompressed pulse-code modulation (PCM) audio as input, accommodating up to 64 channels at 24-bit depth and sample rates up to 192 kHz to handle high-resolution multichannel audio signals. This input format allows for faithful representation of professional-grade audio without any loss of information during compression.^[9]^[1] The overall encoding pipeline starts with lossless matrixing to decorrelate redundancies across channels, followed by linear prediction to model the input audio signals and estimate sample values based on prior samples for identifying remaining redundancies within channels. The differences between actual and predicted values, known as residues, are then calculated to form a compact error signal. These residues undergo entropy coding to minimize the bit representation based on their statistical properties, followed by bitstream formatting that incorporates headers for channel mapping, substream organization, and essential metadata such as sample rate and bit depth. This high-level sequence ensures lossless reconstruction while optimizing storage and transmission efficiency.^[9] The resulting bitstream features a modular structure with sync words at the start of each major frame for precise decoder synchronization, detailed channel assignments that specify the layout and ordering of audio channels, and periodic restart markers that provide error resilience by enabling independent decoding of subsequent segments. In the event of transmission errors, this setup limits impact to within individual packets, typically allowing recovery without audible artifacts or propagation to later frames. The final output is a variable bitrate stream that adapts to the audio content's complexity, commonly delivering compression ratios of 1.5:1 to 2:1 for typical high-resolution sources such as 24-bit/96 kHz multichannel material, thereby reducing data rates by approximately 50% while maintaining bit-perfect fidelity.^[9]

Prediction and Entropy Coding

Meridian Lossless Packing (MLP) employs adaptive linear prediction filters, which can be finite impulse response (FIR) or infinite impulse response (IIR), up to 8th order, to perform intra-channel and inter-channel prediction, exploiting correlations within and across audio channels to reduce redundancy in the linear pulse-code modulation (PCM) signal. The filter coefficients are dynamically adapted using the least mean squares (LMS) algorithm, which iteratively adjusts the parameters to minimize the mean squared prediction error based on recent samples.^[8]^[10] The predicted value for a sample, denoted as \hat{y}(n), is derived from past samples via the linear combination:

\hat{y}(n) = \sum_{k=1}^{P} a_k y(n-k)

where P is the filter order (up to 8), a_k are the LMS-adapted coefficients, and y(n-k) are previous input samples. For inter-channel prediction, the model incorporates samples from adjacent channels to further decorrelate multichannel signals, such as those in surround sound configurations. The residue, representing the irreducible error after prediction, is computed as e(n) = y(n) - \hat{y}(n), ensuring that decoding can recover the exact original sample by reversing the prediction process.^[8]^[10] The residues, which typically follow a Laplacian distribution due to the effectiveness of the prediction stage, are then entropy-coded using Golomb-Rice codes to achieve further compression without loss. Signed residues are encoded by transmitting a sign bit followed by the Golomb-Rice code for the absolute value. The Rice parameter m is adaptively selected for each residue or group of residues based on their statistical properties, such as variance or probability mass function, to optimize the average code length for the observed distribution. For a non-negative integer value x and parameter m, the Golomb-Rice code is formed by encoding \lfloor x / 2^m \rfloor in unary (a run of that many 0s followed by a 1) and appending the m-bit binary representation of x \mod 2^m. This variable-length prefix-free coding assigns shorter codes to more probable small values and longer codes to rarer large values, efficiently packing the sparse residue data.^[10] All operations in MLP's prediction and entropy coding pipeline are fully reversible, with integer arithmetic and exact coefficient quantization preserving bit fidelity throughout encoding and decoding. This design guarantees bit-exact reconstruction of the input PCM stream, independent of platform-specific floating-point variations, thereby maintaining the lossless integrity essential for high-fidelity audio applications.^[8]^[10]

Applications and Formats

Role in DVD-Audio

Meridian Lossless Packing (MLP) serves as the mandatory lossless compression codec for the high-resolution audio layer in the DVD-Audio format, ensuring that all DVD-Audio players include built-in MLP decoders to support playback of high-fidelity content. This requirement stems from the DVD Forum's specification, which designates MLP as the standard method for packing pulse-code modulation (PCM) audio data without any loss in quality, allowing producers to deliver up to 24-bit/96 kHz multi-channel audio while adhering to the format's bit rate limits.^[11] Uncompressed linear PCM (LPCM) is optional but limited by the disc's constraints, making MLP essential for maximizing content duration on standard 4.7 GB single-layer DVDs.^[11] One of the primary benefits of MLP in DVD-Audio is its ability to compress 5.1-channel 24/96 audio, achieving typical reduction ratios of 30 to 50 percent, which enables up to 74 minutes of high-resolution surround sound to fit on a single-layer disc—significantly extending playback time compared to what uncompressed LPCM could achieve under the same capacity and bit rate restrictions.^[12]^[11] This compression fits within the format's maximum audio bit rate of 9.6 Mbps, preventing data overflow while preserving every bit of the original recording.^[11] DVD-Audio discs often incorporate an optional Red Book-compatible stereo layer at 16-bit/44.1 kHz, allowing backward compatibility with standard CD players on hybrid discs, though sample-rate conversion may be needed for seamless integration with the MLP-encoded high-resolution tracks.^[11] MLP also supports embedded metadata in DVD-Audio, including dynamic range control (DRC) through SMART down-mixing coefficients that can be tailored per track to optimize playback across different systems and listener preferences.^[11] For playback, DVD-Audio players perform hardware-based MLP decoding in real time, delivering the uncompressed audio via multi-channel analog outputs or, in later models, bitstream transmission over HDMI or IEEE 1394 (FireWire) interfaces, subject to copy protection mechanisms that initially restricted high-resolution digital outputs to downsampled formats.^[11] This implementation ensures that the full fidelity of MLP-encoded content is accessible only on dedicated DVD-Audio hardware, distinguishing it from DVD-Video players.^[12]

Integration with Blu-ray and Dolby TrueHD

In 2004, Dolby Laboratories developed Dolby TrueHD using the MLP technology licensed from Meridian Audio in 1998, a lossless audio codec that encapsulates the MLP core algorithm with additional Dolby-specific metadata for enhanced functionality.^[12]^[2]^[5] This integration allowed TrueHD to support up to 16 discrete audio channels at resolutions of 24-bit depth and sampling rates up to 192 kHz, though Blu-ray Disc specifications limit practical implementations to eight full-range channels (7.1 surround) at 96 kHz/24-bit or six channels at 192 kHz/24-bit.^[7]^[13] On Blu-ray Discs, Dolby TrueHD serves as an optional high-definition audio format, featuring a mandatory embedded Dolby Digital (AC-3) core track to ensure backward compatibility with legacy devices incapable of lossless decoding.^[14] The MLP core within TrueHD delivers bit-identical lossless reproduction of multi-channel surround sound up to 7.1 configuration, while also accommodating object-based audio extensions such as Dolby Atmos for immersive, height-channel experiences.^[7] This structure enables Blu-ray players to fall back to the lossy core if needed, maintaining playability across a wide range of hardware.^[15] Key enhancements in TrueHD over standalone MLP include robust error detection and concealment mechanisms, which mitigate data loss during transmission or storage without audible artifacts, and hierarchical encoding that prioritizes the AC-3 core for compatibility while layering the lossless MLP data.^[7] These features provide improved resilience in high-bandwidth environments like Blu-ray, contrasting with the more constrained physical media limitations of earlier MLP applications.^[2] Today, Dolby TrueHD remains a cornerstone of Blu-ray audio, widely used in 4K UHD and standard Blu-ray releases for cinematic and music content requiring uncompressed fidelity.^[7] It is commonly bitstreamed via HDMI 2.0 or later interfaces in modern AV receivers and soundbars, ensuring transparent passthrough to compatible decoders without transcoding losses.^[13] While streaming services like Apple Music and Tidal have adopted Dolby Atmos for immersive music playback since around 2020, they typically employ compressed variants such as Dolby Digital Plus rather than full TrueHD, reserving the latter for physical disc-based delivery.^[16]

Performance and Limitations

Compression Efficiency

Meridian Lossless Packing (MLP) achieves typical compression ratios of 1.5:1 to 2:1 for high-resolution multi-channel audio, such as 24-bit/96 kHz six-channel content, reducing data rates from approximately 13.8 Mbps to 4-8 Mbps.^[10] For correlated stereo signals, ratios are lower, around 1.2:1, while uncorrelated noise exhibits near-zero overhead due to the absence of exploitable redundancy.^[10] Efficiency varies with audio characteristics; MLP performs less effectively on low-bit-depth CD audio (16-bit/44.1 kHz stereo), where general-purpose lossless codecs like FLAC or Monkey's Audio often achieve better ratios of 1.6:1 to 1.9:1 by better exploiting limited redundancies in such signals.^[17] In contrast, MLP excels on high-resolution formats due to greater redundancy in dithered signals and inter-channel correlations, enabling it to pack more channels within bandwidth constraints like DVD-Audio's 9.6 Mbps limit.^[10]^[18] Benchmarks from the original MLP encoder demonstrate approximately 50% data reduction on DVD-Audio test sets, such as extending playback from 45 minutes to 100 minutes for uncompressed 24-bit/96 kHz 5.1-channel audio on a single-layer disc.^[10]^[18] Modern implementations, including FFmpeg's MLP encoder, maintain similar ratios while offering improved encoding speeds through optimized prediction and entropy coding processes.^[19] Compared to general-purpose compression like ZIP, MLP is superior for audio-specific redundancies, such as linear prediction across samples and channels, yielding 20-30% better ratios on typical music signals.^[10] However, it is not universally optimal; for instance, TAK can outperform MLP on certain files with high compression modes, particularly low-resolution stereo where TAK's adaptive filters provide tighter packing.^[17]

Implementation Challenges

The original Meridian Lossless Packing (MLP) encoder was computationally intensive, primarily due to the adaptive infinite impulse response (IIR) filters employed in its prediction stage, often requiring several hours to process a single CD-length audio track.^[20] This slowness stemmed from the complexity of optimizing filter coefficients for lossless compression across multiple channels and high-resolution formats, making real-time encoding impractical on hardware of the era. Subsequent open-source implementations, such as the MLP encoder integrated into FFmpeg in 2016, addressed some inefficiencies through algorithmic refinements and parallelization, yet encoding remains CPU-demanding, particularly for multichannel high-bit-depth content. Decoding MLP streams imposes specific hardware and software constraints, as the format was designed for dedicated playback systems. DVD-Audio players were mandated to include MLP decoders to support the standard's high-resolution multichannel audio, ensuring compatibility with discs featuring up to 24-bit/192 kHz content. Similarly, Blu-ray players require MLP decoding capabilities for Dolby TrueHD tracks, which build upon the core MLP algorithm, though not all legacy devices fully support bit-perfect output at maximum resolutions. Software decoding is more limited; while native support exists in players like VLC Media Player, applications such as foobar2000 rely on third-party plugins, like the DVD-Audio Decoder component, to handle MLP files and ensure accurate multichannel playback. MLP's proprietary nature, controlled by Meridian Audio and licensed through Dolby for formats like TrueHD, posed significant compatibility barriers until key patents began expiring in the late 2010s. For instance, the foundational UK patent GB2323754B for IIR-based lossless compression lapsed in February 2017 after its 20-year term, enabling broader open-source adoption without licensing fees.^[8] Prior to this, integration into diverse ecosystems was hindered by legal restrictions, leading to challenges in achieving bit-perfect playback across mixed hardware and software environments, especially for archival or cross-format conversions. Post-expiration, improved interoperability has facilitated tools like FFmpeg for encoding and decoding, though legacy proprietary elements in commercial implementations persist. With the rise of streaming services favoring more versatile open formats like FLAC, MLP's usage has declined significantly since the mid-2010s, as physical media such as DVD-Audio and early Blu-ray discs wane in popularity.^[20] Nonetheless, it remains viable for archival purposes in high-resolution audio preservation, particularly where bit-for-bit fidelity in multichannel setups is essential, supported by ongoing open-source decoders in tools like FFmpeg.