Syncing RiffTrax to Ripped Video (Mac)

Syncing RiffTrax to Ripped Video (Mac)

Requirements

Audacity - for audio editing
ffmpeg - for audio extraction and final construction of the video (installable with brew)
Handbrake - to rip the video from the original media source (if necessary)
Homebrew - to install packages on the Mac
mkvtoolnix - for inspection of media files (installable with brew)

Notes on installation

I’ll be using the movie Gravity as an example, but you should change filenames, accordingly.

As the focus of RiffTrax is really the jokes, we take some liberties with the movie audio - compressing the loud and quiet sections, and mixing down from surround/5.1 to stereo.

`ffmpeg`

For these instructions, ffmpeg needs to be built with the Fraunhofer AAC codec libfdk_aac. Since version 2.0, homebrew no longer allows installation/compilation options for core packages, so this needs to be installed from a third-party tap. On the Mac, this can be done in homebrew using

brew tap homebrew-ffmpeg/ffmpeg
brew install homebrew-ffmpeg/ffmpeg/ffmpeg --with-fdk-aac

`mkvtoolnix`

This package needs to be installed in order to manipulate .mkv files and extract streams. It is available via homebrew:

brew install mkvtoolnix

`srt_presync` and `srt_postsync`

These applications can be installed by cloning this GitHub repository and using pip as follows:

git clone git@github.com:widdowquinn/sync_rifftrax.git
cd sync_rifftrax
pip install .

You can check that the installation completed successfully by issuing, e.g.

% srt_presync --help

 Usage: srt_presync [OPTIONS]

 Entry point for sre_presync

╭─ Options ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ *  --moviepath                 PATH  Path to movie for extraction [required]                                                           │
│    --outpath                   PATH  Directory for output files [default: .]                                                           │
│    --outstem                   TEXT  Stem for output files [default: movie]                                                            │
│    --install-completion              Install completion for the current shell.                                                         │
│    --show-completion                 Show completion for the current shell, to copy it or customize the installation.                  │
│    --help                            Show this message and exit.                                                                       │
╰────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────

Instructions

1. rip the video to `.m4v`

If you need to do this, HandBrake may be useful.

2. extract movie to .mkv (video), .ac3 (audio), and .ass (subtitles)

Use the srt_presync.py script to extract video, audio, and subtitle streams. The command below:

srt_presync.py --moviepath Gravity.m4v --outstem Gravity --outpath movie_export

will extract these files into the movie_export folder, with filestem Gravity.

Note that all subtitle streams will be extracted with names like Gravity_subs2.ass, Gravity_subs3.ass and so on.

5. create new `Audacity` project with movie audio

Open Audacity (this will create a new project)
Import movie audio
Save project

6. compress movie audio

To even up the loud and quiet parts of the movie track, we use compression.

Select the movie audio track
Use Effects->Compressor... to compress the movie audio
The image shown is a first pass at compression. Eventually I settled on a 7:1 ratio as the initial dialogue is so quiet.

7. import RiffTrax audio to the `Audacity` project

Import the appropriate RiffTrax .mp3 file. I’ve found that the US .mp3s work best for blu-ray, and the PAL .mp3s for DVD.
Make sure that the RiffTrax commentary is the lower of the audio tracks - this is important for ducking, later.
From this point on, the RiffTrax README file for the movie is very useful, and it’s handy to have it open in a text editor window

8. trim RiffTrax intro

It can be useful to solo the RiffTrax audio for this stage
Locate the “and we’re back” phrase in the RiffTrax audio, and place the cursor just ahead of that phrase (here, it’s at 2:13.171).
Set the Selection Start manually to zero and cut the section
optionally: create a new track (Tracks->Add New->Mono Track) and paste in the audio you cut, then export the selected audio to a new file (e.g. rifftrax_intro.wav), if you want to preserve it. Then delete that track.

9. initial RiffTrax sync

Using the RiffTrax README (here Gravity_RiffTraxReadme.txt) locate the first Disembaudio line
Here, this is at 1:56.042 (after trimming, it tends to be slightly earlier than the movie time)
Solo the movie audio, and identify the line in the movie - here it is at 2:05.125
Now we can calculate an initial offset for syncing the tracks: 2:05.125 - 1:56.042 = 9.083s.
Select the zero point at the start of the Rifftrax
Add 9.083s of silence at the beginning of the RiffTrax audio using Generate->Silence
Check manually that the audio matches. It can be useful to test the first and last couple of Disembaudio lines, to see if there’s any drift during the track. To correct for noticeable drift, you can delete or insert short periods of silence at the earliest noticeable drift point, to keep the audio in sync.

10. silence Disembaudio

Using the README file, locate and select each Disembaudio line
Silence each selection with Generate->Silence while the line is selected
Export the RiffTrax audio on its own, as a synced track

11. balance riff and movie volume

Play several stretches of movie audio, and note the value in the dB meter when you do so (here, the value was around -10dB for loud sections, -12dB for quiet)
Do the same for the RiffTrax audio (here, the value was around -2dB to -4dB)
Now either increase the volume of the quieter track, or reduce the volume of the louder track, to even out the audio balance, by selecting the track you want to change, and using Effect->Amplify to add or remove gain.
Here I chose to reduce the RiffTrax gain by 6dB, as it was generally louder than the movie, but some sections of the movie audio were already at 0dB, and I wanted to avoid clipping. Play the movie/commentary tracks together to check that the effect is satisfactory (you can always undo).

12. auto-duck the movie

Use Effects->Auto-duck... to quieten the movie track whenever the RiffTrax commentary audio is playing. For this to work, have the movie audio positioned above the RiffTrax audio in the Audacity window.
Select the movie track, then use Effects->Auto-duck... and set the ducking parameters
Playback sections of the audio to check the effect is satisfactory

13. merge the audio and export to `.mp3`

Select both tracks, then use Tracks->Mix and Render
This will generate a single track in the project. Export the audio to an .mp3 file

14. multiplex (mux) the movie and audio

Use srt_postsync.py to compile desired video, audio, and subtitle streams into a new output file. The srt_postsync.py application can combine multiple audio and subtitle streams (with language metadata) if desired, as in the example below. This allows you to keep the commentary and original audio, and any subtitles you wish.

srt_postsync.py --videopath movie_export/Gravity.mkv \
    --audiopaths movie_export/RiffTrax_Gravity.mp3,Gravity.wav \
    --subspaths movie_export/Gravity_subs2.ass,movie_export/Gravity_subs3.ass \
    --audiolangs eng,eng \
    --subslangs eng,fra \
    --title "RiffTrax: Gravity: \
    --outdir riffed_movies \
    --outstem RiffTrax_Gravity

15. check the final output

Play the newly-muxed video in a suitable player and make sure everything’s in sync
If everything seems fine, you’re done!

Summary

srt_presync.py --moviepath Gravity.m4v \
    --outstem Gravity \
    --outpath movie_export
srt_postsync.py --videopath movie_export/Gravity.mkv \
    --audiopaths movie_export/RiffTrax_Gravity.mp3,Gravity.wav \
    --subspaths movie_export/Gravity_subs2.ass,movie_export/Gravity_subs3.ass \
    --audiolangs eng,eng \
    --subslangs eng,fra \
    --title "RiffTrax: Gravity: \
    --outdir riffed_movies \
    --outstem RiffTrax_Gravity

Extras

NTSC to PAL audio tempo change

In some cases, there is no PAL audio available for a RiffTrax commentary (e.g. single-disc Lord of the Rings, Part One). In these cases, it is possible to use the NTSC-synced audio, and speed it up in order to match PAL framerates.

Import the NTSC .mp3 track
In Audacity, use Effects -> Change Tempo and set the percentage change to 4.271 (if you want to convert PAL to NTSC, the change is -4.096)
Click OK

Correcting RiffTrax PAL speed

Some RiffTrax PAL audio (e.g. Return of the King) is not quite synced with PAL DVDs. These often need a small speed-up, with a tempo change of ≈0.125 as above.