Audio Extractor

Audio Extractor

This skill extracts and enhances audio from video files for processing or standalone use.

When to Use This Skill

• Need just the audio from a video
• Want to enhance speech quality
• Preparing audio for transcription
• Creating audio-only content (podcasts, etc.)

What This Skill Does

1. Extract Audio: Separates audio track from video
2. Enhance Quality: Noise reduction, normalization
3. Convert Format: Export as MP3, WAV, AAC, etc.
4. Analyze Audio: Check levels, detect issues

How to Use

Basic Extraction

Extract audio from: ./output/downloads/VIDEO_ID.mp4

With Enhancement

Extract and enhance audio from VIDEO_ID

Specific Format

Extract audio as WAV for transcription

Instructions

Step 1: Analyze Source Audio

ffprobe -v quiet -print_format json -show_streams \
  -select_streams a:0 \
  "./output/downloads/VIDEO_ID.mp4"

Step 2: Extract Audio Track

To MP3 (compressed, good for sharing):

ffmpeg -i "./output/downloads/VIDEO_ID.mp4" \
  -vn -acodec libmp3lame -q:a 2 \
  "./output/downloads/VIDEO_ID.mp3"

To WAV (uncompressed, best for processing):

ffmpeg -i "./output/downloads/VIDEO_ID.mp4" \
  -vn -acodec pcm_s16le -ar 44100 \
  "./temp/VIDEO_ID.wav"

To AAC (high quality, smaller):

ffmpeg -i "./output/downloads/VIDEO_ID.mp4" \
  -vn -acodec aac -b:a 192k \
  "./output/downloads/VIDEO_ID.m4a"

Step 3: Enhance Audio (Optional)

Normalize Volume:

ffmpeg -i "./temp/VIDEO_ID.wav" \
  -af "loudnorm=I=-16:TP=-1.5:LRA=11" \
  "./output/downloads/VIDEO_ID_normalized.mp3"

Reduce Noise + Normalize:

ffmpeg -i "./temp/VIDEO_ID.wav" \
  -af "highpass=f=80,lowpass=f=12000,loudnorm=I=-16:TP=-1.5:LRA=11" \
  "./output/downloads/VIDEO_ID_enhanced.mp3"

Speech Enhancement (for voice clarity):

ffmpeg -i "./temp/VIDEO_ID.wav" \
  -af "highpass=f=100,lowpass=f=8000,loudnorm,compand=attacks=0:points=-80/-80|-45/-15|-27/-9|0/-7|20/-7:gain=5" \
  "./output/downloads/VIDEO_ID_speech.mp3"

Step 4: Report Results

## ✅ Audio Extracted

**Source**: VIDEO_ID.mp4
**Output**: VIDEO_ID.mp3

### Audio Info
- **Duration**: 10:32
- **Sample Rate**: 44.1 kHz
- **Channels**: Stereo
- **Bitrate**: 192 kbps
- **Size**: 12.5 MB

### Quality Assessment
- ✅ Speech clarity: Good
- ✅ Background noise: Minimal
- ✅ Volume levels: Normalized

**Next step**: Use `speech-transcriber` to convert to text

Output

• MP3: ./output/downloads/{video_id}.mp3
• Enhanced: ./output/downloads/{video_id}_enhanced.mp3
• WAV (temp): ./temp/{video_id}.wav

Audio Quality Tips

• Use WAV for transcription (better accuracy)
• Use MP3 for sharing (smaller files)
• Always normalize volume before adding to shorts
• High-pass filter removes rumble, low-pass removes hiss