Video Audio Extraction & Transcoding Guide

When transcribing video files, extracting audio locally before connecting to cloud speech recognition significantly reduces network transfer size and minimizes processing latency.

---

Why Local Audio Extraction?

In video files (such as conference recordings, interviews, webinars, or lecture videos), the visual video streams typically consume over 95% of the total file size. However, speech recognition (ASR) engines only process the audio track.

Metric	Full Video Upload (1 hr 1080p)	Local Audio Extraction (.m4a)	Improvement
File Size	~1.5 GB - 3.0 GB	~25 MB - 40 MB	98% - 99% smaller
Upload Time (50 Mbps)	4 - 8 minutes	4 - 6 seconds	90x faster
Cloud Bandwidth	High data usage	Minimal data usage	Bandwidth friendly

---

Native Format Support (Hardware Accelerated)

DuRT natively leverages Apple's official AVAssetExportSession engine on macOS. When you drag and drop video files in native formats, DuRT automatically and instantaneously extracts a lightweight .m4a audio track locally before dispatching to transcription workflows.

• Natively Supported Formats: .mp4, .mov, .m4v
• Zero Configuration: Built-in Apple Silicon & Intel hardware acceleration, zero external dependencies.
• 100% App Sandbox Compliant: Fully compliant with Apple App Store security policies.

---

Special Video Formats (MKV, FLV, WebM, AVI)

Certain open or legacy video container formats — such as MKV (.mkv), WebM (.webm), FLV (.flv), and AVI (.avi) — are not natively demuxed by macOS system media frameworks.

Sandboxed App Restrictions

DuRT strictly adheres to macOS App Sandbox guidelines for the Mac App Store and does not invoke external command-line utilities directly.

If you import these special formats directly into DuRT:

• The entire original video file will need to be uploaded to cloud recognition servers for remote demuxing;
• Uploading may take longer depending on your network connection speed.

---

Recommended: Fast Local Conversion via FFmpeg

For power users working with MKV, WebM, or other specialized formats, we recommend using the free and open-source FFmpeg tool in your terminal to extract audio or remux prior to importing:

Option 1: Extract Lightweight Audio Track (Recommended)

Extracts an optimized 128 kbps AAC audio file in just 1 to 2 seconds without decoding video:

ffmpeg -i input.mkv -vn -c:a aac -b:a 128k output.m4a

Option 2: Instant Container Remuxing (Zero Quality Loss)

If you prefer keeping the video, you can copy the streams directly into an MP4 container in seconds without re-encoding:

ffmpeg -i input.mkv -c copy output.mp4

After conversion, simply drag the generated .m4a or .mp4 into DuRT to enjoy lightning-fast transcription!