chrisyerga / homerun-dvr

My DIY DVR running on a PC that uses an HDHomeRun device to capture QAM modulated MPEG2 transport streams and extract programs and store them. Integrated with TV guide listings, commercial removal and transcoding.

homerun-dvr

DVR running on a PC that uses an HDHomeRun device to capture QAM modulated MPEG2 transport streams and extract programs and store them. Integrated with TV guide listings, commercial removal and transcoding.

Components

Receiver

Uses the HDHomeRun device which takes a cable TV signal and tuning/filtering parameters to pull a raw MPEG2 transport stream from a given QAM256 channel from your cable provider.

What channels are recorded is controlled by the scheduler informed by TV listings and the interests expressed in the source code parameters. There's no UI to speak of. I have a web app that shows the recording and transcoding pipeline, etc. but it's more for monitoring the system.

With dozens of MPEG transport streams and 100s of programs, the receiver is almost always recording something. It records things opportunistically and then prioritizes transcoding and making them available against a queue full of raw recordings.

MPEG TS Parser

This parser is capable of grabbing a raw MPEG2 transport stream and demuxing it to pull out a given video/audio stream as well as program information.

It parses the PAT and PMT tables to find the elementary streams contained within and can dump those to an abstract VideoStorage which in this implementation stores to a NAT on the local network.

It also parses program information from EIT and PSIP tables, however most of the program information is acquired out-of-band from a TV Listings API service.

Listings

Uses the (now defunct :/) Schedules Direct API to gather EPG listings for the given provider/zip code.

Stitcher

Video channels are constantly recorded and everything from a given channel is stored in a collection of video chunks. The stitcher periodically matches up the Guide listings with the raw video streams from the VideoStorage and stitches them together into video files.

Orchestration / JobQueue

The system can utilize many different computers talking to multiple QAM tuners. There can also be multiple transcoders etc. The code uses both AWS and GCP (nee Azure) to maintain a central Job Queue for recording/transcoding etc.