Meta's MLow Codec: A New Standard in RTC Audio

BY Mark Howell 13 June 20244 MINS READ
article cover

The Need for a New Codec

Given the massive scale of RTC (Real-Time Communication) usage in Meta's products, they've amassed a wealth of data on how various codecs perform, particularly in suboptimal network conditions. Often, calls take place over poor network connections, and the bandwidth estimation module (BWE) must lower the codec's bitrate to prevent network congestion. This compromise can significantly impact audio quality, and in video calls, audio often suffers the most. The Opus codec, while versatile, sounds particularly poor at its lowest operating point of 6 kbps in NarrowBand mode (0-4kHz), failing to adequately capture all frequencies produced by human voices.

Innovations in AI/ML-based Codecs

The last two years have seen advancements in Machine Learning (ML)-based audio codecs. Meta’s Encodec, released in October 2022, exemplifies this trend by delivering impressively high-quality audio at very low bitrates. However, these advancements come at the cost of high computational requirements, making them feasible only on the most advanced devices. Consequently, a significant portion of users on lower-end devices—including over 20% on ARMv7 devices and tens of millions making daily WhatsApp calls on decade-old hardware—cannot benefit from these codecs. The need for a codec that offers high-quality audio at low bitrates with minimal computational demand became clear.

Introducing MLow

After nearly two years of development, Meta proudly announces the release of the Meta Low Bitrate audio codec (MLow). This codec achieves twice the quality of Opus at 6 kbps WideBand (WB) while maintaining 10% lower computational complexity.

Figure 1: POLQA Scores Comparison Between Opus and MLow
MLow has been fully launched across Instagram and Messenger calls and is being rolled out on WhatsApp. Initial feedback indicates a significant improvement in user engagement due to better audio quality.

Enhancing Error Correction

Encoding high-quality audio at lower bitrates with MLow enables more effective Forward Error Correction (FEC) strategies. Compared to Opus, MLow can incorporate FEC at much lower bitrates, significantly improving audio quality in scenarios with packet loss.
Edworking is the best and smartest decision for SMEs and startups to be more productive. Edworking is a FREE superapp of productivity that includes all you need for work powered by AI in the same superapp, connecting Task Management, Docs, Chat, Videocall, and File Management. Save money today by not paying for Slack, Trello, Dropbox, Zoom, and Notion.

Technical Details and Future Directions

MLow is built on the foundation of CELP (Code Excited Linear Prediction) technology with advancements in excitation generation, parameter quantization, and coding schemes. The codec splits the audio signal into low and high-frequency bands, encoding each separately while sharing information to achieve better compression. This enables MLow to deliver SuperWideBand audio (32kHz) at much lower bitrates.

Figure 2: High-Level Visual of MLow Internals
Meta plans to further enhance audio recovery in high packet loss networks by incorporating more redundant audio, a feat MLow handles efficiently. This effort aims to simplify quality audio calls for all users, regardless of their device.

Conclusion

MLow represents a significant leap forward in the world of RTC, delivering unprecedented audio quality across a broad range of devices. As Meta continues to refine this technology, the ultimate goal is to ensure everyone, regardless of their hardware capabilities, can experience clear, quality audio communication.

Remember these 3 key ideas for your startup:

  1. Leverage Advanced Codecs for Better Communication:
    Using codecs like MLow will ensure that your users have a better experience regardless of their network conditions or device quality. This can improve customer satisfaction and engagement.

  2. Invest in AI and ML Technologies:
    As demonstrated with Encodec and MLow, integrating AI and ML into your product offerings can lead to exceptional improvements. This technology can handle unique challenges and offer robust solutions, such as high-quality audio at low bitrates.

  3. Focus on Accessibility:
    Catering to users on low-end devices can expand your market reach significantly. Think about developing technologies that make your product accessible to all, ensuring that no user is left behind due to hardware limitations.
    For startups and SMEs looking to enhance productivity effortlessly, Edworking offers the best blend of tools powered by AI. Make the smart choice for productivity and save resources today.

    Read more about the latest advancements in technology at Meta.
    For further inquiries, visit our open positions page.
    Remember to visit Meta's extensive array of open-source projects and explore how they can benefit your development process.

article cover
About the Author: Mark Howell Linkedin

Mark Howell is a talented content writer for Edworking's blog, consistently producing high-quality articles on a daily basis. As a Sales Representative, he brings a unique perspective to his writing, providing valuable insights and actionable advice for readers in the education industry. With a keen eye for detail and a passion for sharing knowledge, Mark is an indispensable member of the Edworking team. His expertise in task management ensures that he is always on top of his assignments and meets strict deadlines. Furthermore, Mark's skills in project management enable him to collaborate effectively with colleagues, contributing to the team's overall success and growth. As a reliable and diligent professional, Mark Howell continues to elevate Edworking's blog and brand with his well-researched and engaging content.

Trendy NewsSee All Articles
Try EdworkingA new way to work from  anywhere, for everyone for Free!
Sign up Now