If you've ever tried to download a video from Reddit, you've probably ended up with a silent MP4 file. No audio. No error. Just a video that should have sound but doesn't.

This isn't a bug in your downloader. It's how Reddit stores videos.

The Problem

Most video platforms (YouTube, Twitter, etc.) serve videos as a single muxed file — video and audio combined in one stream. Easy to download, plays anywhere.

Reddit doesn't do that. When you upload a video to Reddit, their backend splits it into two separate files stored on v.redd.it:

DASH_720.mp4    ← video only, no audio track
DASH_audio.mp4  ← audio only

When you watch on Reddit, the player loads both files and syncs them client-side. When you download, most tools grab only the video file.

Why It Happens

Reddit uses MPEG-DASH (Dynamic Adaptive Streaming over HTTP). DASH is designed for adaptive streaming where the player picks the best video quality and audio quality independently based on bandwidth.

If you visit a Reddit video URL directly:

https://v.redd.it/abc123/DASH_720.mp4

You'll get a perfectly playable video file — that just happens to have no audio track. The audio lives at:

https://v.redd.it/abc123/DASH_audio.mp4

A naive downloader (curl, wget, basic browser save) only grabs the URL it sees. So you get a silent video.

The Fix

You need to:

1. Download both the video and audio streams
2. Merge them into a single MP4 with FFmpeg

Here's the minimal FFmpeg command that does it:

ffmpeg -i DASH_720.mp4 -i DASH_audio.mp4 \
  -c:v copy -c:a aac \
  output.mp4

The flags matter:

• -c:v copy → don't re-encode video (preserves quality, instant)
• -c:a aac → encode audio as AAC (Reddit's audio is sometimes raw, AAC ensures compatibility)
• Two -i flags → input files; FFmpeg matches them by index

If you skip -c:v copy and let FFmpeg re-encode, you'll lose quality and the operation takes 10x longer.

Doing It Programmatically (Python)

If you're building a tool, yt-dlp handles this automatically when configured correctly:

import yt_dlp

ydl_opts = {
    'format': 'bestvideo+bestaudio/best',
    'merge_output_format': 'mp4',
    'postprocessors': [{
        'key': 'FFmpegVideoConvertor',
        'preferedformat': 'mp4',
    }],
}

with yt_dlp.YoutubeDL(ydl_opts) as ydl:
    ydl.download(['https://reddit.com/r/funny/comments/abc123/title/'])

The key is bestvideo+bestaudio — the + syntax tells yt-dlp to download both streams and merge them. Without the +, you get whatever single stream Reddit returns first (usually video-only).

merge_output_format: 'mp4' ensures the final file is a standard MP4 (FFmpeg might default to MKV otherwise).

Edge Cases

A few things that tripped me up:

1. Some Reddit videos genuinely have no audio. GIF posts and silent screen recordings have no DASH_audio.mp4 file at all. Handle this gracefully:

ydl_opts = {
    'format': 'bestvideo+bestaudio/best',  # falls back to "best" if audio missing
    ...
}

2. Cross-posted videos use different paths. A video cross-posted from r/A to r/B has the original v.redd.it URL. Don't try to construct URLs from the post path — extract the actual v.redd.it URL from the post metadata.

3. NSFW posts require an extra header. Reddit serves NSFW posts to logged-in users, but the video CDN itself doesn't care. You can fetch the video files directly without auth as long as you have the v.redd.it URL.

Why Most Tools Don't Bother

Implementing this correctly requires:

• Detecting that you're on Reddit (URL parsing)
• Extracting the post metadata to find both stream URLs
• Downloading both files (extra bandwidth)
• Running FFmpeg to merge (extra CPU)
• Handling all the edge cases above

A lot of "free Reddit downloaders" skip the merging step because it requires server-side FFmpeg processing or a Wasm FFmpeg in the browser. Both add complexity.

If you want a working version that handles all this, AllClip's Reddit downloader does the merging server-side — paste any Reddit URL and you get an MP4 with audio.