Production Standards

44.1 kHz (Standard Digital Audio) The CD-quality standard. Captures frequencies up to 22.05 kHz — well above the human hearing range (20 Hz – 20 kHz). Use this for:

• Standalone audio ads (no video)
• Podcast ads and audio-only content
• Music tracks and jingles for digital distribution

48 kHz (Video Production Standard) The standard for any audio that will be paired with video. All major video editing software (Premiere Pro, DaVinci Resolve, Final Cut) defaults to 48 kHz. Use this for:

• Social media video ads (Instagram Reels, TikTok, YouTube)
• Television and broadcast spots
• Any audio that will be muxed into a video container

This is the target for all social media platforms, digital display ads, and web-based audio. Why –14 LUFS?

• Spotify normalises to –14 LUFS. YouTube normalises to –14 LUFS. If your audio is louder, the platform will turn it down automatically — and the result often sounds worse than if you had mastered to the target
• –14 LUFS gives your audio breathing room. There is space for dynamic variation, for moments of emphasis, for emotional contrast
• Peak should not exceed –1 dBTP (true peak). This leaves a 1 dB safety margin to prevent clipping on lossy codec conversion (MP3, AAC)

Broadcast standards (EBU R128 for Europe, ATSC A/85 for North America) target –23 LUFS. This is significantly quieter than digital — if you submit a –14 LUFS ad for broadcast, it will be rejected or the broadcaster will compress it aggressively, destroying your careful production.

• Radio spots: master to –23 LUFS integrated
• Podcast pre-rolls distributed via RSS: target –16 to –18 LUFS (closer to podcast norms)
• Television spots: –23 LUFS to –24 LUFS per broadcaster requirements

Platform-Specific Loudness Quirks:

• TikTok: No official normalisation standard. Creators trend louder. Target –12 to –14 LUFS for TikTok specifically
• Instagram Reels: Follows Meta normalisation (–14 LUFS)
• YouTube Shorts: –14 LUFS normalised. Ads louder than –13 LUFS are automatically turned down
• Twitter/X Spaces: Live audio, no normalisation — master to –14 LUFS for consistency

The 8 LU Minimum Rule: Maintain at least 8 LU (Loudness Units) of dynamic range in your final master. This means the loudest moment should be at least 8 LU louder than the average. This gives your audio:

• Natural-sounding voice dynamics (emphasis words are louder, transitions are softer)
• Musical breathing room (crescendos and decrescendos that create emotional movement)
• Listener comfort (the ear does not fatigue from constant loudness)

When More Dynamic Range Is Better:

• Storytelling ads and brand films: 10–14 LU for cinematic feel
• Podcast ads and conversational formats: 8–10 LU for natural speech
• Music-forward ads: 8–12 LU to let the music breathe

When Less Dynamic Range Is Acceptable:

• Short-form social ads (6–15 seconds): 6–8 LU — less time for dynamics to develop
• Noisy-environment ads (transit, outdoor): 6–8 LU — need to cut through ambient noise
• Direct-response radio spots: 6–8 LU — clarity and consistency matter more than nuance

Professional Studio Requirements:

• Noise floor: –60 dB or lower. This means you should hear NOTHING when the voice talent is silent — no air conditioning hum, no computer fan, no traffic rumble
• Room treatment: acoustic panels or foam to eliminate reflections and echo. A voice recorded in an untreated room has a "bathroom" quality that screams amateur
• Microphone: Large-diaphragm condenser for studio (Neumann U87, AKG C414) or dynamic (Shure SM7B) for less-treated spaces. USB microphones are not acceptable for professional advertising
• Pop filter: Always. Plosives (P, B, T sounds) create low-frequency bursts that distort the recording
• Distance: 15–20 cm from microphone. Too close creates proximity effect (boomy bass). Too far captures too much room sound

Remote Recording Specifications: When talent cannot come to a studio, remote recording is acceptable IF:

• Talent uses a professional microphone (not laptop mic or phone)
• Room is treated (closet recording is better than an open room)
• Noise floor is verified before recording (send a 10-second silence test)
• Audio is recorded locally (not through a video call codec — Zoom/Teams audio compression destroys quality)
• ZorgSocial Remote Session Manager handles real-time monitoring and local recording sync

This is the most dangerous misconception. Royalty-free music means you pay a one-time licence fee (instead of ongoing royalties), but:

• You must still purchase the licence
• The licence has territorial restrictions (it may not cover GCC countries)
• The licence has usage restrictions ("social media" may not include paid ads)
• The licence has duration restrictions (1 year? Perpetual? Check the fine print)
• Multiple advertisers may use the same track — your "unique" jingle may appear in a competitor's ad

AI-Generated Music Licensing: Music generated by AI tools (including ZorgSocial Music Generator) has a clearer licensing path:

• ZorgSocial-generated music: full commercial rights included in your subscription. No additional licensing needed. Clear for all territories and platforms
• Third-party AI music (Suno, Udio, etc.): check the platform terms. Some restrict commercial advertising use even on paid plans
• AI-generated music cannot infringe existing copyrights IF the model was trained legally — but this is an evolving legal area. ZorgSocial Music Generator is trained exclusively on licensed datasets

MENA-Specific Licensing Considerations:

• Saudi Arabia and UAE have active copyright enforcement — do not assume "it is not enforced in this region"
• Arabic music rights are often fragmented across multiple publishers and collecting societies
• Traditional Arabic maqam compositions may be in public domain, but specific recordings are not
• Work with a licensing specialist who understands GCC copyright law when using regional Arabic music

The Licensing Checklist (Before Every Deployment):

• Territory: Is the licence valid for every country where the ad will run?
• Platform: Does the licence cover the specific platforms (paid social, programmatic display, streaming audio)?
• Duration: How long can you use the music? Does the licence expire?
• Exclusivity: Is the track exclusive to you, or can competitors use the same music?
• Modifications: Can you edit, remix, or shorten the track? Some licences prohibit modification
• Attribution: Does the licence require a credit or attribution? (Difficult to provide in a 15-second ad)

Common AI Voice Artifacts to Listen For:

• Metallic resonance: A slight ringing or tinny quality, especially on sibilants (S, SH sounds)
• Breathing inconsistency: Either no breathing at all (uncanny) or breathing that sounds pasted in (rhythmically wrong)
• Emphasis errors: Wrong words emphasised in a sentence, or flat delivery on words that should carry emotional weight
• Formant shifts: The voice subtly changes character mid-sentence — sounds like a different person for a fraction of a second
• Plosive handling: P, B, T sounds that either pop excessively or are unnaturally soft
• Sentence boundary artifacts: Slight click or silence gap at the transition between sentences

Arabic AI voice presents unique quality challenges:

• Tashkeel errors: Incorrect vowelisation changes meaning. "كَتَبَ" (wrote) vs. "كُتُب" (books) — the AI must handle diacritics correctly
• Dialect consistency: If the voice starts in Gulf Arabic, it must not drift into Egyptian or Levantine mid-sentence
• Emphatic consonants: Arabic emphatic consonants (ص، ض، ط، ظ) require distinct pharyngealisation that AI sometimes softens
• Connected speech: Arabic connected speech (liaison) patterns differ from English — AI trained primarily on English may produce unnatural Arabic prosody

When to Use Human Voice Instead:

• Emotional storytelling that requires genuine feeling
• Brand spokesperson roles where authenticity is critical
• Regulatory disclaimers in some jurisdictions (check local requirements)
• Premium brand campaigns where any perception of "AI" would damage brand equity

Key Requirements:

• Audibly distinct: The disclaimer must be clearly distinguishable from the main ad content. It cannot be buried under music or sound effects
• Unhurried delivery: Speed-reading disclaimers is a regulatory red flag. The disclaimer must be delivered at a natural, comprehensible pace — typically no faster than 160 words per minute (vs. 140–150 WPM for the main ad content)
• Comparable volume: The disclaimer must be at a volume comparable to the main ad content. Dropping disclaimer volume by 6+ dB (making it half as loud) is a common violation
• Full duration: Do not truncate disclaimers to fit time constraints. If the disclaimer does not fit, the ad is too long — shorten the ad content, not the disclaimer

Industry-Specific Disclaimer Rules:

• Financial services (GCC): Central Bank of UAE (CBUAE) and Saudi Central Bank (SAMA) require specific risk disclosures. In audio, these must be read verbatim, not paraphrased
• Healthcare/Pharma: Side effects and contraindications must be delivered at the same pace and volume as efficacy claims. Rushed disclaimer reading is an FDA/DOH violation
• Real estate: "Prices subject to change" and regulatory authority licence numbers must be included in full
• Food and beverage: Health claims require qualifying statements delivered at comparable prominence

ZorgSocial Compliance Checker for Audio: The Audio Compliance Checker automatically analyses disclaimer segments for:

• WPM (words per minute) comparison vs. main content
• Volume level comparison (LUFS differential)
• Duration adequacy relative to word count
• Language verification (Arabic vs. English disclaimer matches the ad language)

Phone Speaker Limitations:

• Frequency range: Most phone speakers reproduce 200 Hz – 15 kHz effectively. Below 200 Hz (bass) is essentially inaudible. Above 15 kHz (air/shimmer) is reduced
• This means: bass-heavy music beds disappear. Sub-bass rumble that sounded powerful in the studio is simply absent on phone speakers
• Voice clarity is paramount — phone speakers reproduce the 1–4 kHz range (voice presence frequencies) well, so voice-forward mixes translate best

Mobile Mix Tips:

• Boost voice presence (2–4 kHz) by 1–2 dB for phone speaker clarity
• Roll off music bed below 150 Hz — it adds nothing on phone speakers and muddies the voice
• Use sidechain compression to duck the music bed whenever voice is active
• Keep the voice 6–8 dB above the music bed (not 3–4 dB like a music mix)
• Test mono compatibility — phone speakers are mono. Stereo effects that cancel in mono will disappear

Technical Checks:

• Sample rate matches target: 44.1 kHz (audio-only) or 48 kHz (video)
• Bit depth: 16-bit or 24-bit (never 8-bit)
• Integrated loudness: –14 LUFS (digital) or –23 LUFS (broadcast)
• True peak: does not exceed –1 dBTP
• Dynamic range: minimum 8 LU
• No clipping or distortion at any point in the audio
• Fade in/out at start and end (no abrupt cuts that cause pops)

Voice Checks:

• Noise floor: –60 dB or lower (listen to silence sections)
• No mouth clicks, lip smacks, or excessive sibilance
• Pronunciation is correct for all brand names, product names, and technical terms
• Arabic diacritics/tashkeel are correct (for Arabic voiceover)
• Pacing is natural and consistent throughout
• Emphasis on the right words, especially the CTA

Music and SFX Checks:

• Music is properly licensed for territory, platform, and duration
• Music bed does not overpower the voice at any point
• SFX are appropriate volume (complement, not distract)
• No copyright-infringing musical references or samples
• Sonic logo is present and correctly placed (if applicable)

Compliance Checks:

• Disclaimer audio meets regulatory standards (WPM, volume, clarity)
• All required legal disclosures are present and complete
• Language matches target market requirements
• No misleading audio effects (e.g., fake urgency sounds for non-urgent offers)

Mobile Readiness:

• Tested on phone speaker at arm's length
• Voice is intelligible in noisy environment
• Bass-heavy elements are not relied upon for meaning
• Mono compatibility verified (no cancellation issues)

File Delivery:

• File format: WAV (uncompressed) for master, MP3/AAC (320 kbps) for delivery
• File naming convention followed (Brand_Campaign_Version_Date)
• Metadata includes copyright, usage rights, and expiry date
• Backup master file archived in brand asset library