AAF Audio Extractor

WHAT'S NEW IN v1.8

==================

CROSSFADE SUPPORT

-----------------

The extractor now handles crossfades (overlapping clips) correctly.

When two clips overlap on the timeline, their audio is mixed additively

with sample-level clamping, producing smooth transitions without clicks

or digital distortion.

Technical details:

- Additive PCM mixing with per-sample clamping (16/24/32-bit)

- Fast byte copy for non-overlapping regions (performance optimized)

- Chunk scanning to identify overlap boundaries efficiently

- Works for both mono and stereo consolidated tracks

GAIN AUTOMATION

---------------

Audio gain/volume data embedded in AAF metadata is now read and applied:

- Constant gain: Fixed volume level applied uniformly to a clip

  (e.g., -6dB reduction)

- Varying gain (automation): Time-varying volume envelopes with

  keyframe interpolation, applied per-sample across the clip

- Supports AAF ConstantValue, VaryingValue, and Rational parameters

The extraction log shows when gain is detected and applied:

  "Gain detected: N constant, M automated"

  "Gain applied to N clip(s)"

FADE SUPPORT

------------

Linear fade-in and fade-out transitions from the AAF timeline are now

detected and applied:

- Fade-in: Ramps audio from silence to full volume

- Fade-out: Ramps audio from full volume to silence

- Detects transitions adjacent to clips in the AAF sequence

- Works for both simple fades (from/to silence) and crossfades

  (between clips)

SMART TRACK GROUPING

--------------------

Tracks are now automatically classified based on their names and

source file keywords:

- Voice: tracks containing "voice", "dialog", "narr", "vo", or

  named "Audio 1"

- Music: tracks containing "music" or unclassified tracks

- SFX: tracks containing "sfx" or "sound effect"

Each group is exported as a separate file:

  filename_voice.wav

  filename_music.wav

  filename_sfx.wav

L/R channel pairs (e.g., "Audio 2_L" / "Audio 2_R") are

automatically combined into stereo output.

IMPROVED OFFSET WORKAROUND

---------------------------

The experimental workaround for broken AAF exports (where all clips

have start=0) has been improved:

- Now uses the actual timeline edit rate instead of a hardcoded value

- Validates that sufficient audio data is embedded before applying

  sequential offset corrections

- Skips correction when embedded audio is too short, preventing

  incorrect offset assignments

DESCRIPTION

-----------

Tool for extracting audio from AAF (Advanced Authoring Format) files

exported from video editors such as Premiere Pro, DaVinci Resolve, etc.

Automatically extracts:

- Voice track (mono)

- Combined stereo music

REQUIREMENTS

----------

- Windows 10 or higher

- No Python installation or additional dependencies required

HOW TO USE

---------

1. Run AAF_Audio_Extractor.exe

2. Click ‘Add Files’ or ‘Add Folder’ to add AAF files or drag and frop AAF files to the GUI.

3. Select output directory with ‘Browse...’

4. Click ‘Extract All Audio Files’

5. Extracted files will be saved in organised subfolders

SUPPORTED FORMATS

-------------------

- Input: .aaf files

- Output: .wav files (16-bit or 24-bit, depending on the original)

OUTPUT STRUCTURE

--------------------

For each AAF file, a folder is created:

[filename]_extracted/

IMPORTANT NOTES

-----------------

- Fades and automation are NOT applied to the extracted audio

- To export with fades applied, use the export function

of your editor to render the tracks with the automations applied (Premiere Pro, DaVinci, etc.)

- If the AAF file shows warnings about ‘FAT sector count’,

this is normal and safe - the audio will be extracted correctly

SUPPORT

-------

Developed by David García Sánchez

Thanks for supporting this tool's development!