Inside IAMF and ADM: Metadata, Profiles and Rendering

Module G: Standards, agents and QC. Lesson 13 of 20, 30 min.

Abstract

ADM and IAMF describe immersive audio in fundamentally different ways: one is a rich production exchange model, the other a compact distribution bitstream. This lesson dissects both, from audioBlockFormat and chna tables to OBUs, mix presentations and profiles, and shows exactly what survives when an ADM master becomes an IAMF stream.

By the end you can

  • Trace an ADM programme through its content and format elements down to time-varying block formats
  • Parse the axml and chna chunks of a BW64 file and detect cross-reference errors
  • Describe the IAMF v1 OBU structure, audio element types, mix presentations and profiles
  • Plan an ADM-to-IAMF conversion and document precisely which metadata is lost

Every immersive master you deliver is two things at once: audio essence and the metadata that tells a renderer what to do with it. The Audio Definition Model (ADM) and the Immersive Audio Model and Formats (IAMF) describe that metadata in very different ways, and an engineer who understands both can predict what will survive a format conversion. This lesson opens both formats up, down to the chunks, objects and OBUs.

Why metadata internals matter

In the core lesson on Immersive Audio Formats we treated Dolby Atmos ADM BWF and Eclipsa Audio (IAMF) as delivery targets. At the advanced level they are better understood as data models. ADM, standardised by the ITU as Recommendation BS.2076, is a production and exchange model. It is verbose XML that can describe channel beds, moving objects, Ambisonics scenes and binaural signals in one programme. IAMF, published by the Alliance for Open Media (AOMedia), is a distribution model. It is a compact bitstream of Open Bitstream Units (OBUs) built for streaming, playback on very different devices, and fixed decoder complexity.

These two models start from different assumptions, so converting one into the other always involves choices. Some ADM concepts have no direct IAMF equivalent. Some IAMF concepts, such as per-layout loudness information and scalable channel layers, are not native to an ADM master. If a pipeline like the one in The Spatial9 Pipeline produces several formats from one spatial scene, each output must honour the rules of its target model. QC tools, renderers and agents can only check those rules if the engineer can state them.

By the end of this lesson you should be able to:

  • Trace an ADM programme from audioProgramme down to individual audioBlockFormat time segments.
  • Explain how BW64 carries ADM in the axml chunk and links tracks to it through the chna chunk.
  • Describe the IAMF v1 OBU sequence, audio element types, mix presentations and profiles.
  • Reason about what is lost, approximated or preserved when an ADM master becomes an IAMF stream.

The rest of this lesson, its lab and self-check are part of Academy Advanced, a paid course. See who has access.

Key readings

  • ITU-R (2019). Recommendation ITU-R BS.2076-2: Audio Definition Model. International Telecommunication Union.
  • ITU-R (2019). Recommendation ITU-R BS.2088-1: Long-form file format for the international exchange of audio programme materials with metadata. International Telecommunication Union.
  • ITU-R (2019). Recommendation ITU-R BS.2127-0: Audio Definition Model renderer for advanced sound systems. International Telecommunication Union.
  • Alliance for Open Media (2023). Immersive Audio Model and Formats (IAMF) specification. AOMedia.
  • ITU-R (2015). Recommendation ITU-R BS.1770-4: Algorithms to measure audio programme loudness and true-peak audio level. International Telecommunication Union.