Plugins

Back to All Plugins

Segmentarr

Version: 1.5.7 Author: Tw1zT3d2four7 Last Updated: Oct 09 2026, 14:25 UTC

HLS-segmenting stream profile for Dispatcharr: splits XC/URL provider streams into segments, repairs timestamp breaks, and delivers clean MPEG-TS through cvlc to a matching Output Profile.

License: MIT Repository

Downloads

Latest Release

Checksums:

MD5:    b80879859cbc0098a1e6088e75a98574
SHA256: 183da7034a01073b45ad30da5ea7b523ab0b41fbbcc9b469801a40494027ce13

All Versions

Version Download Built Commit MD5 SHA256
1.5.7 Download Oct 09 2026, 14:25 UTC 56c2220 b80879859cbc0098a1e6088e75a98574 183da7034a01073b45ad30da5ea7b523ab0b41fbbcc9b469801a40494027ce13
1.5.6 Download Oct 09 2026, 13:20 UTC 1d958a6 030ac155c0b3067dcbc1c776e1158d03 aee5feff4b265ea7634e2818c0939ce3972e447224b025d896e52df016cdef7c
1.5.3 Download Oct 08 2026, 21:21 UTC 86885f5 acfe8159dc85f145505815f3b994fde4 17f56e46804366451b60859a5460ac3306b87aefe297348f05b2a90202b72bfb

Maintainers: Tw1zT3d2four7 Source: Browse Plugin

Metadata: View full manifest


Plugin README

Segmentarr

Hardware-free stream stabilizer for Dispatcharr. It cuts an unstable IPTV provider stream (Xtream Codes or plain URL) into short HLS segments, repairs timestamp corruption segment by segment, and delivers clean MPEG-TS to Dispatcharr through a final cvlc stage over pipe:1.

provider (XC / URL)
   |  ffmpeg: -c copy -f hls   (keyframe-aligned segments in /dev/shm)
   v
supervisor: validate -> resync -> drop corrupt packets -> stitch PCR/PTS/DTS
   |  ffmpeg finalizer (copy)
   v
cvlc (configurable caching)  ->  pipe:1
   v
Dispatcharr Output Profile (audio stage)  ->  clients

What it fixes

Install

  1. In Dispatcharr, open Plugins, find Segmentarr, and install it.
  2. Open the plugin settings and choose your options.
  3. Press Apply & Synchronize.
  4. Restart any channel that is already playing.

Apply creates a Segmentarr Profile - ... stream profile and a matching Segmentarr Output - ... output profile, and makes both the defaults. Older unlocked Segmentarr profiles are replaced; locked ones are left alone.

Segmentarr and Profilarr both set the same default stream and output profiles. Whichever one you press Apply on last owns them.

Requirements

Settings

Setting Default Effect
Segment Profile Standard (2s) Standard 2s, Low Latency 1s, or Resilient 4s (waits for two segments before starting).
CVLC Network Cache 1000 ms Caching for the final cvlc stage. Off removes cvlc.
Stall Timeout 25s Restart the provider connection if no segment appears for this long.
Max Catch-up Backlog 20s Queued media beyond this is dropped to jump back to live.
Timeline Gap Tolerance 1s Forward PCR jumps up to this are kept; larger breaks are stitched.
Reconnect Delay Ceiling 5s Longest backoff between provider reconnects.
Provider I/O Timeout 15s Provider connection treated as dead after this long without data.
Stream Probe Time 3s How much stream ffmpeg analyses before starting.
Audio Transcoding Override AAC Audio codec applied by the Output Profile (AAC, AC3, E-AC3, Opus, MP3, Copy).

Video and audio are copied in the stream stage; audio is transcoded once, in the Output Profile. Settings are baked into the wrapper scripts at Apply time, so re-apply and restart the channel after changing any of them.

Tradeoffs

Source

github.com/Tw1zT3d2four7/Segmentarr - MIT licensed.