Module 16: Audio Formats & Delivery — Audio Production Foundations

01

Audio Format Fundamentals

Understanding Digital Audio Files

Visual: Comparison diagram of different audio format waveforms and compression

An audio format is how digital audio is encoded, stored, and decoded. Every audio file has:

Container format (file extension)

Codec (encoding method)

Sample rate (samples per second)

Bit depth (precision per sample)

Channels (mono, stereo, surround)

Two Main Categories: 1. Uncompressed/Lossless:

Full quality preserved

Larger file sizes

Examples: WAV, AIFF, FLAC, ALAC

2. Lossy Compressed:

Some data removed (ideally inaudible)

Smaller file sizes

Examples: MP3, AAC, OGG

Why Different Formats Exist:

Production needs full quality (WAV)

Distribution needs small files (MP3)

Archiving needs perfect copies (FLAC)

Each use case has different priorities

The Journey of Audio:

Recording  Production  Master  Distribution  Playback
(High quality) ───────────────────────── (Compressed)

Uncompressed formats store every sample exactly as recorded — no data loss. WAV (Waveform Audio):

Standard for professional audio

Windows native format

Stores raw PCM data

No compression

Large files (~10MB/minute at 44.1/16)

AIFF (Audio Interchange File Format):

Mac equivalent of WAV

Same quality, different container

Used in Logic Pro, older Mac software

Compatible with most pro software

File Size Calculation:

File size = Sample Rate × Bit Depth × Channels × Duration

44,100 × 16 × 2 × 60 seconds = 84,672,000 bits = 10.58 MB per minute

When to Use Uncompressed:

During recording and production

For master delivery

Archiving originals

When quality is priority over size

Limitations:

Large file sizes (storage)

Slow transfers (bandwidth)

Impractical for streaming

Lossless compression reduces file size while preserving every bit of data. FLAC (Free Lossless Audio Codec):

Sample Rate & Bit Depth

The Building Blocks of Digital Audio

Visual: Graph showing sampling and quantization

Sample rate is how many times per second the audio is measured. Common Sample Rates:

Rate	Use Case
44.1 kHz	CD standard, streaming
48 kHz	Video standard, YouTube
88.2 kHz	2× CD rate, some mastering
96 kHz	High-res production
192 kHz	Archival, audiophile

The Nyquist Theorem: To capture a frequency, sample rate must be at least 2× that frequency.

44.1 kHz Can capture up to ~22 kHz (human hearing limit)

48 kHz Can capture up to ~24 kHz

Why 44.1 kHz is Standard:

Human hearing: 20 Hz - 20 kHz

44.1 kHz captures all audible frequencies

CD standard since 1982

Efficient balance of quality/size

Higher Sample Rates:

More headroom for processing

Better for time-stretching/pitch-shifting

Some believe they sound 'better'

But no audible frequencies above 20 kHz

Recommendation:

Record/produce at 48 kHz or 96 kHz

Deliver at whatever platform requires

44.1 kHz for audio-only, 48 kHz for video

Bit depth is the precision of each sample — how accurately amplitude is recorded. Common Bit Depths:

Depth	Dynamic Range	Use
16-bit	96 dB	CD, final delivery
24-bit	144 dB	Recording, production
32-bit float	1528 dB	Processing headroom

What Bit Depth Affects:

Dynamic range (loud to quiet)

Noise floor (quietest possible sound)

Precision of quiet sounds

16-bit Reality:

96 dB dynamic range

More than enough for playback

Human hearing: ~120 dB range

Quiet room: ~30 dB ambient

Effective range needed: ~90 dB

24-bit for Recording:

Headroom for levels

Clean gain reduction

Better for processing

Convert to 16-bit for delivery

32-bit Float:

Virtually unlimited headroom

Can clip in DAW and recover

Processing format, not delivery

No additional audible benefit over 24-bit

Dithering adds noise to prevent distortion when reducing bit depth. The Problem: Converting 24-bit to 16-bit means losing precision. Without dither, this causes quantization distortion (especially on quiet sounds). The Solution: Add very low-level noise (dither) that masks the quantization error. Sounds counterintuitive, but noise is better than distortion. When to Dither:

Only when reducing bit depth

Only on the FINAL bounce

24-bit master 16-bit delivery

Not needed going up (1624)

Not needed staying same (2424)

Dither Types:

Type	Character
POW-r 1	Subtle, neutral
POW-r 2	More high-frequency
POW-r 3	Maximum noise shaping
MBIT+	iZotope standard
Rectangular	Basic, flat noise

Dither Settings:

Use noise shaping for less audible noise

Apply at final export only

Don't dither twice (cumulative noise)

Practical Rule: Bouncing 24-bit mix to 16-bit for CD? Dither Bouncing 24-bit mix to streaming platforms? Check their specs (often 24-bit accepted)

Converting between sample rates requires careful processing. Why Convert:

Recorded at 96 kHz, need 44.1 kHz for CD

Video at 48 kHz, audio at 44.1 kHz

Platform requires specific rate

Conversion Quality Matters: Poor conversion creates artifacts. Use quality converters. Simple Ratios (Easier):

88.2 44.1 (÷2)

96 48 (÷2)

Simple division, cleaner

Complex Ratios (Harder):

48 44.1 (not simple ratio)

Requires interpolation

Quality depends on algorithm

Best Practices:

Work at target rate if possible

Use high-quality SRC (iZotope, Weiss)

Simple ratios when possible

Don't convert repeatedly (quality loss)

Quality Converters:

iZotope RX/Ozone (excellent)

Weiss Saracon (mastering standard)

SOX (free, high quality)

Most DAWs have decent built-in

Workflow Recommendation:

Record at 48 kHz (video compatible)

Or record at 96 kHz and convert

Deliver at platform requirements

Archive at original rate

03

Streaming Platform Standards

What Each Platform Requires

Visual: Comparison chart of platform specifications

Spotify's requirements and how they process your audio. Upload Specifications:

Format: WAV or FLAC preferred

Sample rate: 44.1 kHz (recommended)

Bit depth: 16-bit or 24-bit

LUFS target: -14 LUFS

How Spotify Processes:

Transcodes to OGG Vorbis (free) or AAC (premium)

Normalizes to -14 LUFS

Quality tiers: 96/160/320 kbps

Loudness Normalization:

If your master is:
-8 LUFS  Turned DOWN 6 dB
-14 LUFS  No change
-18 LUFS  Turned UP 4 dB

Why -14 LUFS Matters:

Masters louder than -14 get turned down

No benefit to over-compression

Louder ≠ louder on Spotify

More dynamic masters sound better

Spotify Loud Mode:

Some users disable normalization

Your loud master plays at full level

But most users have it on

Recommendation: Master to -14 LUFS integrated with good dynamics. Over-compressed masters lose punch when normalized.

Apple Music supports high-resolution and Dolby Atmos. Upload Specifications:

Format: WAV or AIFF via aggregator

Sample rate: 44.1-192 kHz (hi-res supported)

Bit depth: 16-24 bit

LUFS target: -16 LUFS (recommendation)

Apple Digital Masters:

Certification program for quality

Requires 24-bit, 96 kHz minimum

Master specifically for Apple's codec

Apple-approved mastering tools

How Apple Processes:

Transcodes to AAC 256 kbps (standard)

Or streams lossless (ALAC) if enabled

Spatial Audio (Dolby Atmos) supported

Loudness Normalization:

Sound Check normalizes to -16 LUFS

Can be disabled by users

More conservative than Spotify

Lossless Tier:

ALAC up to 24-bit/192 kHz

Requires user opt-in

Growing audiophile market

Recommendation: Provide highest quality possible. Apple accepts up to 192 kHz/24-bit. If creating Apple Digital Master, follow their specific guidelines.

YouTube is primarily video, but audio specs matter. Upload Specifications:

Format: Audio in video container (MP4, MOV)

Sample rate: 48 kHz (video standard)

Bit depth: 16-bit or higher

Audio codec: AAC or PCM

How YouTube Processes:

Re-encodes all audio to AAC/Opus

Normalizes loudness (varies)

Quality depends on video quality selected

Audio Quality Tiers: | Video Quality | Audio | |---------------|-------| | 360p | 128 kbps AAC | | 720p | 192 kbps AAC | | 1080p+ | 256 kbps AAC | | Premium | Opus (better) | Loudness on YouTube:

Normalization targets ~-14 LUFS

But implementation varies

Over-compressed audio can distort after encoding

For Music Videos:

Upload at 1080p or higher (better audio codec)

48 kHz sample rate (match video)

Avoid clipping — YouTube encoding is harsh

Leave 1-2 dB headroom

YouTube Music:

Pulls from YouTube uploads

Quality depends on original upload

No separate audio-only path

Other major platforms and their specifications. Tidal:

Supports MQA, lossless, hi-res

Upload WAV 44.1-96 kHz

Masters at -14 LUFS

Audiophile focus

Amazon Music:

HD tier: 16-bit/44.1 kHz

Ultra HD: 24-bit/192 kHz

Normalizes to -14 LUFS

Upload via aggregator

SoundCloud:

Transcodes to 128 kbps MP3 (free)

Or 256 kbps AAC (Go+)

No loudness normalization

Upload highest quality source

Bandcamp:

Lossless download available

Upload WAV/FLAC

No transcoding for paid downloads

Artists control format options

Aggregator Requirements: Most distributors (DistroKid, CD Baby, TuneCore) accept:

WAV or FLAC

44.1 kHz/16-bit minimum

96 kHz/24-bit maximum (varies)

Stereo

Universal Master: Create one master at:

44.1 kHz or 48 kHz

24-bit (dither to 16-bit if needed)

-14 LUFS integrated

True peak < -1 dBTP

This works for most platforms.

04

Loudness Measurement Deep Dive

Peak, RMS, and LUFS — What They Measure and When to Use Each

Visual: Comparison of Peak, RMS, and LUFS measurement methods

Peak measurement shows the maximum instantaneous signal level — critical for preventing clipping. What Peak Measures:

The highest point the waveform reaches

Instantaneous — a single moment in time

Shows clipping risk, not loudness

Types of Peak:

Type	What It Measures	Use Case
Sample Peak	Highest digital sample value	Basic metering
True Peak	Analog signal between samples	Streaming masters

The True Peak Problem: Digital audio is made of discrete samples. The analog signal reconstructed between samples can exceed those sample values.

Sample values: 0dB, -2dB, 0dB
True peak: +0.8dB (clipping!)

Peak Targets: | Context | Maximum | |---------|----------| | Recording | -6 dBFS (headroom) | | Mixing | -3 to -6 dBFS | | CD Master | -0.3 dBTP | | Streaming | -1.0 dBTP | | Broadcast | -2.0 dBTP | Key Insight: Peak tells you if you'll clip — not how loud something sounds. A snare hit might peak at 0dB but sound quieter than sustained synth at -6dB.

RMS (Root Mean Square) measures average signal energy — closer to perceived loudness than peak. What RMS Measures:

Average power over a time window

Mathematically: √(sum of squared samples / count)

Smooths out peaks and valleys

Better loudness indicator than peak

Why RMS Matters:

Signal A: Peak 0dB, RMS -18dB (very dynamic)
Signal B: Peak 0dB, RMS -8dB (compressed)
Both peak at 0dB, but B sounds MUCH louder

Crest Factor (Peak-to-RMS Ratio):

Content Type	Typical Crest Factor
Classical music	18-25 dB
Acoustic/Jazz	14-18 dB
Rock/Pop	10-14 dB
EDM/Compressed	6-10 dB
Brickwall limited	3-6 dB

Crest Factor Tells You:

How dynamic the signal is

Low crest = heavily compressed

High crest = lots of dynamics

Limitations of RMS:

No frequency weighting

Treats bass and treble equally

Doesn't match human hearing perception

That's why LUFS was created...

LUFS (Loudness Units Full Scale) measures perceived loudness using psychoacoustic modeling. Why LUFS Exists: RMS treats all frequencies equally. But humans don't hear that way:

We're most sensitive to 2-4 kHz (speech range)

Bass sounds quieter than it measures

LUFS applies K-weighting to match human perception

K-Weighting:

Reduces low frequencies (below 200 Hz)

Slightly boosts 2-4 kHz region

Models how humans perceive loudness

International standard (ITU-R BS.1770)

Three LUFS Measurements:

Type	Window	Use Case
Momentary	400ms	Real-time feedback
Short-term	3 sec	Section comparison
Integrated	Full length	Delivery target

LUFS vs. LKFS: They're identical. LUFS is EBU term (Europe), LKFS is ITU term (international). Loudness Range (LRA):

Measures dynamic variation in LUFS

High LRA = more dynamics

Low LRA = more compressed

Typical pop: 5-10 LRA

Classical: 15-20+ LRA

All three measurements serve different purposes — professionals use them together. Quick Reference:

Measurement	Question It Answers
Peak	Will it clip?
RMS	How much average energy?
LUFS	How loud does it sound?

Same Track, Different Readings:

Example: Heavily compressed EDM track
Peak: -0.3 dB (near maximum)
RMS: -8 dB (high average energy)
LUFS: -7 LUFS (very loud)
Crest: 7 dB (low dynamics)
Example: Dynamic orchestral piece  
Peak: -0.5 dB (near maximum)
RMS: -22 dB (low average energy)
LUFS: -18 LUFS (moderate perceived)
Crest: 21 dB (high dynamics)

Workflow Integration: Recording: Watch peak (stay below -6 dB) Mixing: Watch RMS for balance, peak for headroom Mastering: Target LUFS, verify true peak Delivery: Check integrated LUFS matches platform target Pro Tip: If your LUFS is high but your track sounds weak, you probably have too much sub-bass energy (contributes to RMS but LUFS de-emphasizes it). K-weighting reveals the truth!

05

Loudness Standards

LUFS, True Peak, and Loudness Normalization

Visual: LUFS meter showing integrated, short-term, and momentary readings

LUFS (Loudness Units Full Scale) measures perceived loudness, not peak level. LUFS vs. dB Peak:

dB Peak: Maximum signal level (can clip)

LUFS: How loud it SOUNDS to humans

A quiet song can have high peaks

A loud song can have few peaks

LUFS Measurement Types:

Type	Measures	Use
Momentary	Last 400ms	Real-time feedback
Short-term	Last 3 seconds	Section loudness
Integrated	Entire song	Overall loudness

Target LUFS by Platform: | Platform | Target | |----------|--------| | Spotify | -14 LUFS | | Apple Music | -16 LUFS | | YouTube | -14 LUFS | | Broadcast (US) | -24 LUFS | | CD (no normalization) | -9 to -6 LUFS | Why LUFS Replaced Peak:

Peak doesn't correlate with perceived loudness

LUFS uses psychoacoustic modeling

Better represents listener experience

Enables meaningful normalization

True Peak measures actual maximum level between samples — critical for encoding. The Problem with Sample Peaks: Digital meters show sample values. But the analog reconstruction between samples can exceed those values.

Sample Peak: 0 dB (looks fine)
True Peak: +0.8 dB (actually clipping!)

Why True Peak Matters:

Lossy encoding creates new peaks

DACs reconstruct analog signal

Peaks between samples cause distortion

Streaming platforms will clip

True Peak Targets: | Context | True Peak Max | |---------|---------------| | Streaming master | -1.0 dBTP | | Broadcast | -2.0 dBTP | | CD master | -0.3 dBTP | | Safe for all | -1.5 dBTP | Measuring True Peak:

Regular peak meters don't show it

Use true peak (TP) meters

Most loudness meters include TP

Check before final bounce

Avoiding True Peak Issues:

Leave headroom (-1 dB or more)

Use limiter with true peak mode

Check true peak on final file

iZotope, FabFilter, stock limiters have TP

Streaming platforms adjust playback volume — the loudness war is over. How Normalization Works:

Platform measures your master's LUFS

Compares to their target (-14 LUFS Spotify)

Adjusts playback gain to match

All songs play at similar loudness

The Death of the Loudness War:

Master A: -8 LUFS Turned down 6 dB Master B: -14 LUFS No change Master C: -18 LUFS Turned up 4 dB

Result: All play at same perceived loudness

Why Over-Compression Now Hurts:

Hyper-compressed master gets turned down

Loses impact, sounds flat

Dynamic master plays at same loudness

But retains punch and energy

Loudness Range (LRA):

Measures dynamic variation

Higher LRA = more dynamics preserved

Pop/rock: 5-8 LRA typical

Classical: 15-20+ LRA

Modern Mastering Philosophy:

Master for the song, not loudness

-14 LUFS is guidance, not law

Some songs should be -12, some -16

Dynamics serve the music

How to use loudness meters in practice. Essential Meters:

LUFS meter (integrated)

True peak meter

Short-term LUFS

Loudness range (LRA)

Metering Plugins:

Youlean Loudness Meter (free, excellent)

iZotope Insight (comprehensive)

NUGEN VisLM (broadcast standard)

Stock meters (most DAWs include)

Workflow:

Mix to sound good (don't watch meters)

Check integrated LUFS at end

Check true peak

Adjust limiter if needed

Re-check measurements

Reading the Meter:

Integrated LUFS: -13.5 LUFS ✓ (close to -14)
True Peak: -0.4 dBTP ✗ (too hot)
Loudness Range: 7 LRA ✓ (good dynamics)

Common Fixes: | Problem | Fix | |---------|-----| | Too loud (LUFS) | Less limiting, more dynamics | | Too quiet (LUFS) | More limiting (careful!) | | True peak over -1 | Ceiling lower on limiter | | Low LRA | Less compression, more dynamics | Final Check: Bounce the file, re-import, measure. Export process can change levels slightly.

06

Export & Delivery Workflow

From Session to Final File

Visual: Flowchart showing export workflow

Pre-export checklist to ensure quality delivery. Before Bouncing: Session Check:

[ ] All tracks at correct levels?

[ ] Automation playing correctly?

[ ] No muted/soloed tracks by accident?

[ ] Correct bus routing?

[ ] Master bus processing engaged?

Quality Check:

[ ] No clipping on any channel?

[ ] True peak below -1 dBTP?

[ ] LUFS in appropriate range?

[ ] Listen in mono — any issues?

[ ] Fade ins/outs applied?

Technical Check:

[ ] Correct sample rate for destination?

[ ] Correct bit depth?

[ ] Offline bounce (not real-time)?

[ ] Dither applied (if reducing bit depth)?

The Final Listen: Before exporting, play the entire song:

No distractions

At moderate volume

Note any issues

Fix before bounce

Don't:

Export without listening first

Assume it sounds right

Skip the pre-export check

Understanding export settings in your DAW. Essential Settings: File Format:

WAV for masters and production

AIFF for Mac-centric workflows

MP3/AAC for previews only

Sample Rate:

Match session rate (ideal)

Or convert to delivery rate

44.1 kHz for audio-only

48 kHz for video

Bit Depth:

24-bit for masters

16-bit for CD (with dither)

32-bit float if staying in DAW

Bounce Range:

Set locators precisely

Include any reverb tails!

Leave small silence at start (0.5-1s)

Leave decay at end

Offline vs. Real-Time:

Offline	Real-Time
Faster	Actual playback
Consistent	Can catch issues
Recommended	Use for checking plugins

File Naming: Use consistent naming:

Artist_SongTitle_Version_Date.wav
MyBand_MySong_Master_20240315.wav

Different destinations need different versions. Master Versions to Create: 1. Production Master (Archive):

Highest quality available

Session sample rate/bit depth

No limiting on master bus

For future re-mastering

2. Distribution Master:

44.1 kHz or 48 kHz / 24-bit

Mastered with limiting

-14 LUFS typical

For streaming/aggregators

3. CD Master (if needed):

44.1 kHz / 16-bit

With dither

Red Book compliant

For physical CDs

4. Video Master (if needed):

48 kHz (video standard)

24-bit

May need specific loudness

5. Preview/Reference:

MP3 320 kbps or AAC 256 kbps

For sending to artists/labels

Lower quality, small files

Folder Structure:

/ProjectName/
  /Masters/
    Production_Master.wav
    Distribution_Master.wav
    CD_Master.wav
  /Previews/
    Preview.mp3

Verify your exports before sending anywhere. Post-Export QC: 1. Re-Import and Listen:

Import bounced file back into DAW

A/B with session (match levels!)

Listen for artifacts, clicks, dropouts

2. Check File Info:

Correct sample rate?

Correct bit depth?

Correct duration?

Metadata intact?

3. Measure Again:

Check LUFS of bounced file

Check true peak

Should match session

4. Play on Different Systems:

Headphones

Car (if possible)

Phone speaker

Different speakers

Common Export Problems:

Problem	Cause	Fix
Clicks at start/end	No fade	Add short fades
Cut off reverb	Range too short	Extend bounce range
Distortion	True peak clipping	Lower limiter ceiling
Wrong format	Settings error	Re-export
Missing elements	Muted tracks	Check session

The "Coffee Test": After exporting, step away. Come back with fresh ears. Listen again. Still sound right?

07

Metadata & Tagging

Identifying Your Audio Files

Visual: Metadata editor showing various tag fields

Metadata is data about your data — information embedded in audio files. Why Metadata Matters:

Identifies your music

Enables discovery

Tracks royalties

Organizes libraries

Required for distribution

Common Metadata Fields:

Field	Purpose
Title	Song name
Artist	Performer name
Album	Collection name
Year	Release year
Genre	Musical category
Track #	Position in album
Album Art	Visual representation
ISRC	Unique recording identifier
Comment	Notes, credits

Format Support:

MP3: ID3v2 tags (extensive)

FLAC: Vorbis comments (good)

WAV: Limited metadata (INFO chunks)

AIFF: ID3v2 supported

Metadata Persists: Once embedded, metadata travels with the file. Tag it once, identified forever.

These tags are required for professional distribution. Mandatory for Streaming: ISRC (International Standard Recording Code):

Unique 12-character code per recording

Format: CC-XXX-YY-NNNNN

Required for royalty tracking

Get from label or distributor

UPC/EAN (Release Barcode):

For albums/EPs

Required for physical and digital

12 or 13 digit code

Basic Tags:

Title: My Song Name
Artist: My Artist Name
Album: My Album Name
Year: 2024
Genre: Electronic (match streaming categories)
Track Number: 01

Extended Tags:

Composer (for publishing royalties)

Lyricist (if different from composer)

Publisher

Record Label

Copyright notice

Consistency is Key:

Spell artist name identically everywhere

Use exact same titles across platforms

Consistent genre classification

This helps algorithms find you

Album art is visual metadata — it's essential for recognition. Technical Requirements:

Spec	Requirement
Minimum size	1400×1400 px (most platforms)
Recommended	3000×3000 px
Format	JPG or PNG
Color space	RGB, sRGB
File size	Under 10 MB
Shape	Square (1:1 ratio)

Quality Guidelines:

High resolution, no pixelation

No watermarks or logos (unless part of art)

No URLs or promotional text

No social media handles

Legible at small sizes (thumbnail)

Platform-Specific:

Spotify requires centered content (may be cropped)

Apple uses rounded corners

YouTube displays wide, may crop

Embedding Art:

Embed directly in audio file

Also provide separate file

Most aggregators require both

Common Rejection Reasons:

Too low resolution

Contains text that violates policies

Copyright infringement (stock photos)

Not square aspect ratio

Poor image quality

Software for adding and editing metadata. Desktop Applications: Mp3tag (Windows/Mac):

Industry standard

Free

Batch editing

Auto-tagging from databases

Supports all formats

Kid3 (Cross-platform):

Free, open source

Clean interface

Good format support

Yate (Mac):

Powerful automation

Script support

Professional features

Paid

MusicBrainz Picard:

Free

Auto-tags from database

Great for tagging existing libraries

Online Databases:

MusicBrainz (open database)

Discogs (physical releases)

Gracenote (industry standard)

DAW Metadata: Some DAWs support embedding metadata during export:

Logic Pro: File > Bounce > Metadata tab

Cubase: Audio Mixdown settings

Pro Tools: Bounce to Disk settings

Best Practice:

Tag files immediately after mastering

Include all essential fields

Verify before sending to distributor

Keep backup with metadata intact

08

Common Delivery Scenarios

What to Send and When

Visual: Decision tree for delivery format selection

Preparing files for streaming platforms via aggregators. Aggregators (Distributors):

DistroKid, CD Baby, TuneCore, etc.

They deliver your music to streaming platforms

Each has slightly different requirements

Standard Requirements:

Format: WAV or FLAC
Sample Rate: 44.1 kHz (some accept up to 96 kHz)
Bit Depth: 16-bit or 24-bit
Channels: Stereo
Loudness: -14 LUFS recommended
True Peak: < -1 dBTP

Preparing Your Release:

Final master at 44.1/24-bit

Include all metadata

Album art at 3000×3000

ISRC codes (if you have them)

Release date planned

What Aggregators Do:

Check file quality

Transcode for each platform

Distribute metadata

Collect royalties

Report streams

Delivery Checklist:

[ ] Master file(s) in correct format

[ ] All metadata filled out

[ ] Album artwork

[ ] ISRC codes

[ ] Release info (date, genre, etc.)

[ ] Payment info set up

Preparing music for film, TV, advertising — specific requirements. What Sync Clients Need: Full Mix:

WAV, 48 kHz/24-bit (video standard)

True peak: -3 dBTP (conservative)

Leaves room for dialogue

Stems/Alternate Mixes:

Instrumental version (no vocals)

Acapella (vocals only)

Full mix

Sometimes: Drums only, no drums, etc.

Common Stem Breakdown:

- Full Mix
Instrumental (no vocals)
Vocals Only
Music (no drums or bass)
Drums Only
Bass Only

File Naming for Sync:

SongTitle_FullMix_48k24b.wav
SongTitle_Instrumental_48k24b.wav
SongTitle_Vocals_48k24b.wav

Cue Sheets:

Document song use in production

Required for royalty collection

PRO/performance rights tracking

Technical Standards:

48 kHz sample rate (matches video)

Broadcast loudness varies by region

Leave headroom for post-production

No watermarks or artifacts

Preparing masters for physical formats — vinyl, CD, cassette. Vinyl Mastering: Special Considerations:

Bass must be mono (center)

Limited high-frequency energy

Less compression is better

Shorter sides = louder cut

De-essing may be needed

Different master than digital

Technical Limits: | Factor | Limit | |--------|-------| | Side length | ~20-22 min (loud) | | Bass | Mono below 300Hz | | High frequencies | Careful above 15kHz | | Sibilance | Problematic | What to Send to Cutting Engineer:

High-resolution master (96/24 if available)

Ideally a vinyl-specific master

Track sequence and gaps

Side A/B split

CD Mastering:

44.1 kHz / 16-bit (Red Book)

Dither when converting

DDP file for replication

PQ codes for track markers

Cassette:

Similar to vinyl considerations

Even more limited frequency response

Often separate master needed

Preserving your audio for the future. Archive Format Recommendations: Session Archive:

Collect All/Consolidate session

Include all audio files

Save at highest resolution

Document plugin settings

Master Archive:

Original session sample rate/bit depth

No limiting/processing on master bus

Pre-master mix for future re-mastering

Final masters in multiple formats

Backup Strategy:

Local Drive  External Drive  Cloud
                              
  Working      Backup 1     Off-site

3-2-1 Rule:

3 copies of your data

2 different storage types

1 off-site location

Storage Formats:

WAV/AIFF for audio (standard)

FLAC for space efficiency (lossless)

Never archive as MP3/AAC (lossy)

Metadata for Archives:

Date created/modified

Project notes

Credits

Technical specifications

Version history

Long-Term Considerations:

Formats change over time

Standard formats (WAV) will persist

Update storage media every 5-10 years

Test backups periodically

A/D

Analog Approach

Analog delivery meant physical formats: vinyl, tape, CD. Each had inherent technical limitations (vinyl frequency response, tape hiss) that required specific mastering approaches. Distribution required manufacturing, shipping, and retail. The listener received exactly what was pressed/recorded.

Digital Approach

Digital delivery democratized distribution but introduced complexity: multiple formats, codecs, loudness normalization, platform-specific transcoding. A single master may be heard as OGG on Spotify, AAC on Apple, and various YouTube qualities. The creator doesn't fully control the final playback quality.

Key Difference

Analog was 'what you master is what they hear.' Digital is 'what you master is transcoded, normalized, and streamed based on bandwidth.' Modern mastering must anticipate how platforms will process audio, leaving headroom and avoiding over-compression that sounds worse after normalization.

SCI

The Psychoacoustics Behind Lossy Compression

Lossy audio compression exploits psychoacoustic masking: our ears' inability to hear certain sounds. Simultaneous masking means a loud sound covers a quieter sound at nearby frequencies — the encoder discards the masked sound. Temporal masking means sounds immediately before or after a loud sound are also inaudible. By analyzing these perceptual limits, codecs like MP3 and AAC remove data that we theoretically can't hear. Higher bitrates keep more detail, reducing the chance of audible artifacts.

EX

Bob Ludwig / Mastering Engineer

Engineer: Gateway Mastering

Bob Ludwig creates different masters for different formats: streaming masters at -14 LUFS with dynamics, CD masters optimized for that format, and vinyl masters with mono bass and controlled highs. Shows that one master doesn't fit all.

Spotify 'Loud & Clear'

Engineer: Spotify Engineering

Spotify's loudness normalization changed mastering practices industry-wide. Before normalization, masters competed for loudness. After, dynamics became viable again because all songs play at similar levels regardless of master loudness.

Neil Young / Pono Player

Engineer: Various

Neil Young's Pono project championed high-resolution audio (24/192), arguing lossy formats removed musical information. While commercially unsuccessful, it highlighted the quality debate and pushed platforms toward lossless options.

Taylor Swift / 1989 (Taylor's Version)

Engineer: Various

Re-recording catalog demonstrates how mastering changes with times. New versions are mastered differently for streaming era — different loudness targets, different dynamic range — compared to original 2014 masters.

History of Audio Formats

Engineer: Industry Evolution

1877: Edison's phonograph (cylinder). 1948: Columbia's LP (33⅓ RPM vinyl). 1963: Philips Compact Cassette. 1982: Sony/Philips CD (44.1kHz/16-bit). 1999: Napster and MP3 revolution. 2015: Streaming becomes dominant. Each format balanced quality, convenience, and economics differently.

ERR

!

Over-compressing for loudness when streaming will normalize anyway

Why it's bad: Streaming platforms turn down loud masters. Over-compressed audio loses dynamics and punch but doesn't play any louder. A -8 LUFS master gets turned down 6dB on Spotify.

Better approach: Master to around -14 LUFS with good dynamics. Your music will play at the same loudness as heavily compressed masters but will sound more dynamic and punchy.

!

Ignoring true peak, only watching sample peaks

Why it's bad: True peaks between samples can exceed 0dB even when sample peaks show headroom. Streaming encoding creates new peaks. Result: distortion on playback systems.

Better approach: Use true peak limiting with ceiling at -1 dBTP or lower. Measure true peak on the final bounce, not just during mixing. Leave headroom for encoding.

!

Converting sample rates repeatedly

Why it's bad: Each sample rate conversion introduces potential artifacts. Converting 4844.14844.1 degrades quality. Non-integer ratio conversions are particularly problematic.

Better approach: Work at target rate when possible. If conversion needed, do it once with a quality converter. Keep original files at native rate for archives.

!

Sending MP3s to streaming aggregators

Why it's bad: Aggregators transcode to their own formats. Starting with MP3 means: lossy encoding second lossy encoding. Quality degrades significantly. Artifacts become obvious.

Better approach: Always send WAV or FLAC to aggregators. They'll create the compressed versions. Starting lossless ensures the best possible quality after their transcoding.

!

Not dithering when reducing bit depth

Why it's bad: Converting 24-bit to 16-bit without dither causes quantization distortion, especially audible on quiet passages, reverb tails, and fade outs. The distortion is harsh and obvious.

Better approach: Apply dither when bouncing from 24-bit session to 16-bit delivery. Use noise-shaped dither for least audible result. Only dither once, on the final bounce.

!

Not embedding metadata before distribution

Why it's bad: Without metadata, your song is unidentified in databases. Royalties may be missed. Songs display incorrectly on players. Professional releases always include proper tags.

Better approach: Tag all files with complete metadata (title, artist, ISRC, etc.) before sending to aggregators. Use consistent spelling. Include album art. Verify before submission.

EAR

01

Codec Comparison

How to: Encode the same master as WAV, FLAC, MP3 320, MP3 192, MP3 128. Import all into a session. Level-match and A/B compare.

What to notice: At what bitrate do you hear degradation? Focus on cymbals, reverb tails, and sibilance. These reveal codec artifacts first. MP3 128 should be noticeably worse than WAV.

02

Loudness Normalization Experience

How to: Create two masters of the same song: one at -14 LUFS with dynamics, one at -8 LUFS heavily compressed. Play through Spotify (normalize on). Compare.

What to notice: They should play at similar loudness. But the dynamic version should sound more punchy and musical. The compressed version sounds flat and lifeless at the same playback level.

03

True Peak vs Sample Peak

How to: Set a limiter ceiling at 0dB (sample peaks). Bounce. Check with a true peak meter. Then set limiter to -1dBTP (true peak mode) and compare.

What to notice: The 0dB sample peak version may show true peaks above 0dB. This would clip after encoding. The -1dBTP version stays clean through the entire chain.

04

Dither Audibility

How to: Create two versions: 24-bit to 16-bit with dither, and without dither. Boost quiet sections by 40dB. Listen to the noise floor and very quiet sounds.

What to notice: Without dither, you'll hear harsh quantization distortion on quiet sounds and fades. With dither, you hear soft noise but no distortion. Dither is always better.

05

Sample Rate Conversion Quality

How to: Convert a 96kHz file to 44.1kHz using different software/algorithms. Compare the results with the original (played at 44.1kHz).

What to notice: Quality converters should sound nearly identical to the original. Poor converters may add ringing, artifacts, or change the high-frequency character.

Q

Q1

What is the recommended LUFS target for Spotify?

Show Answer

Q2

Why is true peak limiting important for streaming delivery?

Show Answer

Q3

When should dithering be applied?

Show Answer

Q4

What is the difference between LUFS and dB peak?

Show Answer

Q5

What format should you send to streaming aggregators?

Show Answer

Q6

Why do vinyl masters need mono bass?

Show Answer

Q7

What is an ISRC code?

Show Answer

Q8

What sample rate is standard for video/film work?

Show Answer

KEY

Takeaway 1

Takeaway 2

Takeaway 3

Takeaway 4

Takeaway 5

Takeaway 6

Takeaway 7

Takeaway 8

Audio Formats & Delivery

Core Concepts

Audio Format Fundamentals

Sample Rate & Bit Depth

Streaming Platform Standards

Loudness Measurement Deep Dive

Loudness Standards

Export & Delivery Workflow

Metadata & Tagging

Common Delivery Scenarios

Analog vs Digital

Analog Approach

Digital Approach

Key Difference

The Science

The Psychoacoustics Behind Lossy Compression

Real-World Examples

Bob Ludwig / Mastering Engineer

Spotify 'Loud & Clear'

Neil Young / Pono Player

Taylor Swift / 1989 (Taylor's Version)

History of Audio Formats

Common Mistakes

Over-compressing for loudness when streaming will normalize anyway

Ignoring true peak, only watching sample peaks

Converting sample rates repeatedly

Sending MP3s to streaming aggregators

Not dithering when reducing bit depth

Not embedding metadata before distribution

Listening Tests

Codec Comparison

Loudness Normalization Experience

True Peak vs Sample Peak

Dither Audibility

Sample Rate Conversion Quality

Self-Assessment

Key Takeaways