xHE-AAC Audio
xHE-AAC combines Extended HE-AAC based on USAC with MPEG-D loudness and dynamic-range control for speech and music delivery.
A podcast alternates spoken narration, music, and both together. At a tight delivery bitrate, a codec needs to handle the changes without the music collapsing when speech-oriented assumptions stop fitting.
xHE-AAC combines Extended HE-AAC based on Unified Speech and Audio Coding (USAC) with MPEG-D loudness and dynamic-range control. Its parts address the coded sound, switching among delivery versions, and playback level.
Match the signal's character
Speech has predictable pitch and short-term spectral structure. Music and mixed content can require a broader representation of transients, tones, and textures.
USAC includes predictive speech-oriented and transform-based coding tools. The encoder chooses an appropriate representation and signals what the decoder must reconstruct. This unification lets a programme move between voices and general audio within the format rather than requiring the listener to select a different codec for every segment.
The reconstruction still represents a lossy approximation. Its efficiency is useful because a limited budget can be distributed using tools suited to different signal behavior.
Adapt delivery without changing the timeline
An adaptive service can prepare several encoded representations of the same programme, then select among them as the connection permits. xHE-AAC's bitstream tools support suitable switching points and continuous reconstruction across prepared changes.
| Delivery condition | Service action |
|---|---|
| More data budget available | Select a higher-rate prepared version |
| A restricted connection | Select a lower-rate prepared version |
| A switch point | Continue the same programme timeline with the selected representation |
The decoder must initialize or update its coding state appropriately at a switch. The service must also align versions and supply the needed configuration. “Adaptive” therefore describes a complete prepared delivery process, not a stored file rewriting its own audio.
Use metadata to control the listening level
MPEG-D loudness and dynamic-range control supplies information for consistent loudness and suitable level-dependent processing.
For an illustrative normalization calculation, if a programme's measured loudness is 5 dB below a chosen playback target, a simple gain correction is +5 dB. The amplitude multiplier is 10^(5/20), about 1.778.
That also raises peaks, so loudness gain, available headroom, and dynamic-range processing need to cooperate. The purpose is to make programmes usable across listening situations without treating each encoded sample value as if it directly specified room volume.
The loudness/control system is a defined part of xHE-AAC delivery, rather than a promise that any app reading an audio file will apply identical gain.
Identify the actual bitstream
USAC content is not an ordinary AAC-LC stream with some new tags. It requires its own reconstruction tools. The xHE-AAC profile also provides earlier AAC-family decoding capability, but an older LC or HE-AAC decoder does not thereby gain USAC support.
For an Emby item containing xHE-AAC, distinguish whether the client receives that source representation or a server-made compatible version. This identifies which decoder job is being performed, while the actual player's loudness/output processing determines how the decoded programme reaches the listening system.