Understanding the Core Differences Between AES67 and AVB

Audio-over-IP technology has fundamentally reshaped professional audio workflows, offering unprecedented flexibility, scalability, and integration with existing IT infrastructure. Among the many protocols available, AES67 and Audio Video Bridging (AVB) stand out as two of the most widely deployed for high-performance audio transport. While both aim to deliver reliable, low-latency streaming over Ethernet, they take fundamentally different approaches to clocking, network management, and interoperability. Choosing the right protocol for your installation requires a clear understanding of these differences, the specific demands of your application, and the capabilities of your network hardware.

What Is AES67?

AES67 is an open, industry-standard protocol developed by the Audio Engineering Society. Published in 2013 and updated to AES67-2018, it defines a set of interoperability requirements for high-performance audio-over-IP streaming. The standard specifies audio sample rates (44.1, 48, 96 kHz), bit depths (16, 24, 32), packet formats (L24, L16, AM824), and timing using the Precision Time Protocol (PTPv2, IEEE 1588-2008). Critically, AES67 does not mandate any particular transport mechanism; it works over standard Ethernet networks using RTP (Real-time Transport Protocol) and provides mechanisms for device discovery and connection management.

Because AES67 is an open standard, it is designed to bridge the gap between proprietary ecosystems like Dante, Livewire, Ravenna, and Q-LAN. Any device that conforms to AES67 can communicate with any other AES67-compliant device, regardless of the underlying proprietary protocol. This makes AES67 a natural choice for facilities that need to integrate equipment from multiple manufacturers without vendor lock-in.

Key Characteristics of AES67

  • Open Standard: Published by AES, freely available, and supported by numerous manufacturers.
  • PTP-Based Clocking: Uses PTPv2 (IEEE 1588-2008) for synchronization, allowing sub-microsecond jitter.
  • Standard Ethernet: Runs on off-the-shelf switches without requiring managed QoS (though QoS is recommended).
  • Low Latency: Typically achievable in the range of 1–10 ms, depending on packet size and network loading.
  • High Channel Count: Up to 64 channels per stream at 48 kHz/24-bit, and more with next-generation variants.
  • No Built-in Redundancy: Redundancy must be handled at the network level (e.g., ST 2022-7 seamless switching).

AES67 is widely adopted in broadcast, live sound, and installed sound applications. It is particularly strong in environments where interoperability with legacy systems or multi-vendor setups is critical. For example, a radio station using a Dante-based console can send AES67 streams to a Ravenna-based processor without additional converters, provided both devices support AES67.

What Is AVB?

Audio Video Bridging (AVB) is a set of IEEE standards (802.1BA, 802.1Qat, 802.1Qav, 802.1AS, etc.) that together provide deterministic, low-cost, real-time audio and video transport over Ethernet. AVB was initially developed by the IEEE 802.1 Audio/Video Bridging Task Group and is now maintained under the Time-Sensitive Networking (TSN) umbrella. Unlike AES67, AVB is a complete, integrated solution that defines everything from network topology to clock synchronization to stream reservation.

AVB relies on managed network switches that support AVB-specific features: credit-based shaper (CBS) for traffic prioritization, stream reservation protocol (SRP) for bandwidth guarantees, and generalized precision time protocol (gPTP) for clock distribution. AVB networks require a domain with at least one grandmaster clock (GM) and all switches and endpoints compliant with IEEE 802.1AS. This strict hardware dependency ensures predictable, sub-2ms latency end-to-end across seven hops, making AVB ideal for performance-critical applications like live concert sound, post-production suites, and automotive infotainment.

Key Characteristics of AVB

  • IEEE Standards Suite: Covering clocking (802.1AS), stream reservation (802.1Qat), traffic shaping (802.1Qav), and network bridging (802.1BA).
  • Deterministic Latency: Guaranteed maximum latency of 2 ms over seven network hops for Class A traffic.
  • Stream Reservation: Bandwidth is reserved end-to-end; if insufficient bandwidth exists, the connection is refused.
  • Managed Switches Required: Only switches that support AVB/TSN can participate in an AVB domain.
  • Plug-and-Play: Devices automatically discover each other and establish streams without manual IP configuration.
  • Built-in Redundancy: Can be achieved through redundant network paths (e.g., dual-homed endpoints).

AVB is commonly used in fixed installations where dedicated, managed network hardware is justified. Typical examples include concert halls with distributed amplifiers, automotive audio networks, and high-end recording studios that require deterministic performance across many channels. AVB is also the basis for the Milan protocol, which adds a control layer and device profiles for studio use.

Performance Comparison

When comparing AES67 and AVB, performance must be evaluated across several dimensions: interoperability, latency, timing accuracy, network requirements, scalability, and total cost of ownership. Each protocol excels in different areas, and the optimal choice depends on the specific application.

Interoperability and Ecosystem Support

AES67 was expressly designed for interoperability. Because it is an open standard, it enjoys broad support across the professional audio industry. Major manufacturers like Audinate (Dante), Riedel, Lawo, and Yamaha all offer AES67-compatible interfaces. This means an audio engineer can connect a Dante-equipped mixing console to a Ravenna-based processing engine without proprietary bridging hardware. However, AES67 does not define a unified control plane; each ecosystem retains its own discovery and routing methods, so integration often requires manual configuration or middleware.

AVB is also based on open IEEE standards, but its strict hardware requirements limit interoperability at the network level. An AVB stream can only travel through AVB-capable switches and endpoints. While AVB endpoints from different manufacturers can communicate if they implement the same IEEE standards (e.g., Milan), the switch infrastructure must be AVB-compliant. This reduces flexibility compared to AES67, which can run on any network that supports PTP and RTP. In practice, AVB is more common in closed, proprietary ecosystems like automotive or in specialized AVB/Milan installations.

Latency and Determinism

Both protocols offer very low latency, but their approaches differ. AES67 latency is configurable via packet size and network settings. Typical real-world figures range from 1 ms (with very small packets) to 10 ms (with larger packets). Because AES67 uses RTP over UDP, it is subject to network congestion and does not inherently guarantee bandwidth. However, with managed switches and proper Quality of Service (QoS) configuration, AES67 can achieve round-trip latencies under 5 ms.

AVB is built for determinism. Using the credit-based shaper (CBS) and stream reservation, AVB guarantees that audio traffic will not be delayed by non-time-sensitive traffic. Class A traffic (highest priority) has a maximum latency of 2 ms across seven hops. This makes AVB ideal for live performance where consistent, predictable timing is critical—for example, in-ear monitor systems where even 1 ms of jitter can be audible.

Clocking and Synchronization

Both protocols rely on Precision Time Protocol variants. AES67 uses PTPv2 (IEEE 1588-2008) in a profile that requires at least one grandmaster clock. The profile is defined in AES67-2018 and is compatible with SMPTE ST 2059-1/2 for broadcast synchronization. AES67 can achieve clock accuracy within ±1 μs.

AVB uses gPTP (IEEE 802.1AS), which is a profile of IEEE 1588 optimized for audio/video applications. gPTP simplifies the grandmaster election and ensures sub-microsecond synchronization across the entire domain. Because gPTP is embedded in the network switch hardware, AVB can achieve tighter synchronization than AES67 in large, multi-hop networks. For applications requiring super-low jitter (e.g., high-end analog-to-digital conversion), AVB’s hardware-assisted clocking provides an advantage.

Network Requirements and Infrastructure Cost

One of the most significant practical differences is in network hardware.

AES67 can operate on standard, unmanaged switches, although QoS is strongly recommended. Many installations use Layer 2 managed switches with DiffServ markings to prioritize audio packets. AES67 does not require any specific switch capabilities beyond IGMP snooping for multicast traffic. This makes AES67 very attractive for budget-conscious or retrofit projects where upgrading to managed switches is not feasible.

AVB mandates AVB-capable switches. These switches are more expensive than standard managed switches because they must implement the CBS, SRP, and gPTP hardware. In a large installation, the cost of the switch infrastructure can be a significant portion of the budget. However, because AVB guarantees bandwidth and latency, the network is easier to design and more reliable for real-time traffic. For new builds where the network can be specified from scratch, the additional cost may be justified by the simplified commissioning and guaranteed performance.

Scalability and Channel Count

AES67 supports up to 64 channels per stream at 48 kHz/24-bit in its basic format. Higher channel counts (e.g., 512 channels) are possible with proprietary extensions like Dante’s 128-channel streams. AES67 can be scaled in bandwidth by adding more streams and using multicast or unicast. The total number of channels is limited only by network capacity.

AVB typically supports up to 256 channels per stream in its current implementations (based on the Milan standard). Because AVB reserves bandwidth, scaling the network requires careful planning to ensure all stream reservations fit within the switch’s available capacity. In practice, AVB is often used for smaller to medium channel counts (e.g., 64–128 channels) in critical applications, while AES67 is favored for very large systems (e.g., broadcast plants with thousands of channels).

Redundancy and Reliability

AES67 does not define a native redundancy mechanism. However, network-level redundancy can be implemented using SMPTE ST 2022-7, which sends duplicate streams over two separate network paths. Many AES67-compatible devices support this feature. Additionally, standard network redundancy protocols (e.g., RSTP, MRP) can be used, though they may introduce failover delays.

AVB supports built-in redundancy through dual-homed endpoints and redundant network paths using the Stream Reservation Protocol (SRP). AVB’s deterministic nature also makes it easier to predict behavior during failover. For mission-critical live sound, AVB with redundant networks offers higher reliability out of the box.

Which Protocol Is Better for Your Setup?

The answer depends on your technical requirements, budget, and infrastructure. Below is a practical decision framework.

Choose AES67 if:

  • You need to integrate equipment from multiple manufacturers or existing proprietary systems.
  • Your network infrastructure already exists and you cannot replace all switches with AVB-capable units.
  • You require very high channel counts (e.g., 512+) or flexible routing configurations.
  • Your budget is limited, and you prefer to use standard IT networking hardware.
  • You work in broadcasting, post-production, or corporate AV where multicast and PTP are already familiar.

Choose AVB if:

  • You are designing a new installation from the ground up and can specify AVB-capable switches.
  • Your application demands guaranteed, sub-2ms latency with zero jitter—for example, live performance or in-ear monitoring.
  • You want plug-and-play device discovery and automatic stream configuration without manual IP addressing.
  • You need built-in redundancy without relying on external protocols.
  • You work in automotive, live sound touring, or high-end studio environments where deterministic performance outweighs cost.

Hybrid Approaches and Emerging Standards

In many modern installations, AES67 and AVB are not mutually exclusive. For example, a studio might use AES67 to connect Dante and Ravenna devices while using AVB (via Milan) for critical microphone preamps and monitors. Bridges exist—such as the Luminex network switches that can translate between AES67 and AVB domains, or software-based gateways. Additionally, the emerging SMPTE ST 2110 standard for broadcast uses AES67 for audio and is widely adopted in television production. Meanwhile, TSN (Time-Sensitive Networking) is evolving to offer even more advanced deterministic capabilities that may blur the line between AES67 and AVB in the future.

Practical Implementation Considerations

Regardless of your protocol choice, proper network design is essential. For AES67, ensure that PTP is correctly configured with a reliable grandmaster clock, and apply QoS mapping (DSCP 44-46 for audio). For AVB, verify that all switches are from the same vendor and firmware generation to avoid interoperability issues. Test your network with a tool like Wireshark to monitor PTP traffic, stream reservations, and latency.

Always involve your IT team early—audio-over-IP protocols place demands on the network that typical office traffic does not. Switches must be configured with multicast filtering, bandwidth limits, and security settings to prevent unauthorized streams. For AVB, ensuring that the switch topology does not exceed the maximum hop count (typically 7) is critical.

Future Outlook

The trend in professional audio is toward greater convergence. AES67 continues to be the glue that connects diverse ecosystems, while AVB/Milan is carving out a niche in high-performance installations. With the ongoing development of TSN, we may see a single unified standard that combines the flexibility of AES67 with the determinism of AVB. Until then, understanding the strengths and weaknesses of each protocol will remain essential for system designers.

Conclusion

Both AES67 and AVB offer robust audio-over-IP transport, but they serve different priorities. AES67 champions open interoperability and works over existing networks, making it the pragmatic choice for multi-vendor environments and cost-sensitive projects. AVB delivers unmatched determinism and ease of setup in greenfield installations where performance is paramount. By evaluating your specific latency requirements, network hardware, interoperability needs, and budget, you can confidently select the protocol that best supports your professional audio setup.

For further reading, consult the AES67-2018 standard, the IEEE AVB Task Group page, and an Audinate guide to AES67 and Dante integration.