Radio/audio broadcast format support and loudness requirements

[bokelley]

Parent: #1919
Related: #2037 (broadcast TV spot formats), #2036 (Ad-ID)

Problem

The protocol has broadcast TV spot formats but no radio/audio broadcast formats. Radio broadcast shares structural similarities with TV broadcast (no impression pixels, physical delivery, Ad-ID, measurement windows) but differs in creative formats and loudness standards.

Additionally, the audio asset requirements schema (audio-asset-requirements.json) has no loudness fields at all. The audio reference formats (audio_standard_15s, audio_standard_30s, audio_standard_60s) don't declare loudness requirements. This means a creative agent has no schema-level guidance on loudness for any audio format.

Loudness Standards (from Sam Sousa / AES TD1008)

Per AES Technical Document TD1008 v3.13, recommended loudness targets vary by distribution:

Context	Target LUFS	Tolerance	Standard
OTA Broadcast TV	-24	±2 dB	ATSC A/85
Audio ads / interstitials	-18	±2 dB	AES TD1008
Streaming/SSAI audio	-16 to -18	Platform-specific	Varies
Podcast	-16	±1.5 dB	Spotify/Apple guidance

Key insight from Sam: -24 LUFS is correct for OTA broadcast TV but too quiet for audio-only and digital distribution. Creative reuse across channels (a spot made for TV also running on streaming audio) means the source should be at -18 LUFS. Playout systems normalize anyway, but starting louder prevents quality loss from normalization up.

Greg confirmed: CTV and broadcast TV both manage audio levels at playout. But SSAI for live streaming handles it in the delivery stack, not always at playout — making source loudness more important.

Proposal

1. Add loudness fields to audio-asset-requirements.json

"loudness_lufs": {
  "type": "number",
  "description": "Target integrated loudness in LUFS (e.g., -18 for audio ads per AES TD1008, -24 for broadcast TV per ATSC A/85)"
},
"loudness_tolerance_db": {
  "type": "number",
  "minimum": 0,
  "description": "Acceptable deviation from loudness_lufs target in dB"
},
"true_peak_dbfs": {
  "type": "number",
  "description": "Maximum true peak level in dBFS (e.g., -1 for audio, -2 for broadcast TV)"
}

2. Add loudness to audio reference formats

Update audio_standard_15s, audio_standard_30s, audio_standard_60s with loudness_lufs: -18 per AES TD1008.

3. Radio broadcast formats

Add reference formats for radio broadcast spots, following the same pattern as TV broadcast:

• :15, :30, :60 radio spots
• Audio file only (MP3/AAC/WAV), no VAST/DAAST, no impression trackers
• Ad-ID support (same industry_identifiers field)
• Loudness at -18 LUFS per AES TD1008
• 48kHz stereo, minimum 192kbps

4. Radio broadcast measurement

Radio measurement has its own characteristics (different from TV):

• Nielsen Audio (PPM panels in top markets, diary in smaller markets)
• Measurement windows may differ from TV's C3/C7 model
• Needs input from radio sellers on how measurement data arrives

5. Documentation

• Add docs/creative/channels/radio.mdx (or extend existing audio docs)
• Add loudness guidance table to docs/creative/channels/audio.mdx and docs/creative/channels/video.mdx referencing AES TD1008

Open questions

• Should radio broadcast be a separate channel doc or a section within the existing audio.mdx?
• Do radio stations use Ad-ID or a different creative identifier system?
• What does radio measurement data look like? (Nielsen Audio PPM vs diary)
• Should we define separate loudness profiles by distribution context (OTA broadcast, streaming SSAI, podcast DAI)?

References

• AES TD1008 v3.13 — Audio loudness for interstitials/advertisements
• ATSC A/85 — US broadcast TV loudness standard
• EBU R128 — European broadcast loudness standard