Skip to main content

​
Overview

FFmpeg can be used for silence detection and removal in audio and video processing. This guide shows you how to:
  • Detect silence in media files and export metadata
  • Remove silence from media files automatically

​
Silence Detection with Metadata Export

The silencedetect filter identifies silent segments in audio based on noise threshold and duration. By combining it with ametadata, you can export detailed metadata about when silence occurs in your media files.

​
FFmpeg Command for Silence Detection

ffmpeg -i https://storage.rendi.dev/sample/popeye_talking.mp4 -af "silencedetect=noise=-20dB:d=0.5,ametadata=mode=print:file=silence_metadata.txt" -f null -

​
Command Breakdown:

  • -i input.mp4 - Input video or audio file
  • -af - Audio filter
  • silencedetect=noise=-20dB:d=0.5 - Detects silence below -20dB lasting at least 0.5 seconds
    • noise=-20dB - Threshold for silence (adjust based on your audio; -20dB is moderate, -30dB is more sensitive)
    • d=0.5 - Minimum duration of silence in seconds
  • ametadata=mode=print:file=silence_metadata.txt - Exports metadata to a file
    • mode=print - Prints metadata for each frame
    • file=silence_metadata.txt - Output file for metadata
  • -f null - - No video output (we’re only analyzing)

​
Run Silence Detection in Rendi

To run this command in Rendi, you need to use the /v1/run-ffmpeg-command endpoint: 
curl --location 'https://api.rendi.dev/v1/run-ffmpeg-command' \
--header 'X-API-KEY: <api-key>' \
--header 'Content-Type: application/json; charset=utf-8' \
--data '{
    "input_files": {
        "in_1": "https://storage.rendi.dev/sample/popeye_talking.mp4"
    },
    "output_files": {
        "out_1": "silence_metadata.txt"
    },
    "ffmpeg_command": "-i {{in_1}} -af \"silencedetect=noise=-20dB:d=0.5,ametadata=mode=print:file={{out_1}}\" -f null -"
}'
When commands’ status reaches SUCCESS, you’ll get the metadata file in the response: 
{
    "command_id": "089dd36c-723c-4a0a-b68a-8e8cbcc1afd2",
    "status": "SUCCESS",
    "output_files": {
        "out_1": {
            "file_id": "da09eaa7-904f-45e2-a727-74760b2696f6",
            "storage_url": "https://storage.rendi.dev/files/.../silence_metadata.txt",
            "status": "STORED"
        }
    }
}
The metadata file will contain timestamps of all detected silence segments, which you can parse for further processing.

​
Silence Removal

While FFmpeg includes a silenceremove filter, the recommended method for removing silence is the Jump Cuts technique, which involves a two-step process: Step 1: Detect Silence and Get Timestamps First, use silencedetect to identify where silence occurs in your video: 
curl --location 'https://api.rendi.dev/v1/run-ffmpeg-command' \
--header 'X-API-KEY: <api-key>' \
--header 'Content-Type: application/json; charset=utf-8' \
--data '{
    "input_files": {
        "in_1": "https://storage.rendi.dev/sample/popeye_talking.mp4"
    },
    "output_files": {
        "out_1": "silence_metadata.txt"
    },
    "ffmpeg_command": "-i {{in_1}} -af \"silencedetect=noise=-20dB:d=0.5,ametadata=mode=print:file={{out_1}}\" -f null -"
}'
Step 2: Parse Timestamps and Create Jump Cuts Parse the silence timestamps from the metadata file, then use them to create precise jump cuts that completely remove the silent segments. This gives you:
  • Complete removal of silence (not just audio compression)
  • Maintained video/audio sync
  • Precise control over what gets cut
  • Shorter final video duration

​
Example: Running Jump Cuts in Rendi

After analyzing the silence metadata, you can create jump cuts to keep only the non-silent portions. For example, to keep segments from 0.0-5.7s, 11.0-18.0s, and 19.0-20.0s: Base FFmpeg Command: 
ffmpeg -i https://storage.rendi.dev/sample/popeye_talking.mp4 -vf "select='between(t,0.0,5.7)+between(t,11.0,18.0)+between(t,19.0,20.0)',setpts=N/FRAME_RATE/TB" -af "aselect='between(t,0.0,5.7)+between(t,11.0,18.0)+between(t,19.0,20.0)',asetpts=N/SR/TB" popeye_jumpcuts.mp4
Run in Rendi: 
curl --location 'https://api.rendi.dev/v1/run-ffmpeg-command' \
--header 'X-API-KEY: <api-key>' \
--header 'Content-Type: application/json; charset=utf-8' \
--data '{
    "input_files": {
        "in_1": "https://storage.rendi.dev/sample/popeye_talking.mp4"
    },
    "output_files": {
        "out_1": "popeye_jumpcuts.mp4"
    },
    "ffmpeg_command": "-i {{in_1}} -vf \"select='\''between(t,0.0,5.7)+between(t,11.0,18.0)+between(t,19.0,20.0)'\'',setpts=N/FRAME_RATE/TB\" -af \"aselect='\''between(t,0.0,5.7)+between(t,11.0,18.0)+between(t,19.0,20.0)'\'',asetpts=N/SR/TB\" {{out_1}}"
}'
Command Breakdown:
  • -vf "select='...'" - Video filter to select specific time ranges
  • between(t,0.0,5.7)+between(t,11.0,18.0)+between(t,19.0,20.0) - Keep frames between these timestamps
  • setpts=N/FRAME_RATE/TB - Reset presentation timestamps for smooth playback
  • -af "aselect='...'" - Same selection applied to audio
  • asetpts=N/SR/TB - Reset audio timestamps to maintain sync
For detailed examples of creating jump cuts with specific timestamps, see the FFmpeg Cheatsheet - Jump Cuts section.
This two-step approach is superior to silenceremove for video files because it actually cuts out the silent portions entirely, making your videos shorter and more engaging, rather than just modifying the audio track.

​
Use Cases

Silence Detection with Metadata:
  • Analyze podcast episodes for silence patterns
  • Quality control for audio recordings
  • Identify timestamps for manual editing
  • Generate subtitles timing
Silence Removal:
  • Clean up podcast recordings
  • Speed up video tutorials by removing pauses
  • Create tighter, more engaging content
  • Reduce file size by removing dead air

​
Tips for Best Results

  1. Test with different thresholds: Start with -20dB and adjust based on your audio quality
  2. Consider minimum duration: Use d=0.5 (0.5 seconds) to avoid removing natural pauses in speech
  3. Preview before batch processing: Test on a small sample first
  4. Use metadata for precision: Detect silence first, then use timestamps for precise cuts
  5. Quality encoding: When re-encoding, use -c:a aac -b:a 128k for good audio quality