BackBeat

BackBeat batch-downloads and post-processes YouTube music videos for use as background videos in rhythm games such as YARG.

It reads one or more CSV playlists, lets you choose what to process in a small Tkinter setup window, then downloads video-only sources with yt-dlp and converts them into game-friendly output files.

What BackBeat Does

Given a CSV playlist, BackBeat can:

• download the selected YouTube video
• strip audio so the output is video-only
• optionally remove black bars with cropdetect
• apply a start delay or a black pre-roll pad
• apply playback speed correction
• pad the final result to a 16:9 canvas without scaling the original image
• encode to webm or mp4
• skip rows that were already processed with identical CSV parameters

The included backbeat.csv covers a large Rock Band music-video set, so a full run can take a long time depending on connection speed and CPU performance.

Sync Accuracy

BackBeat's timing values are practical approximations, not guarantees. Official music videos often use different edits, mixes, intros, or lengths than the in-game stems they are meant to accompany.

If you find a better source video, a more accurate delay or speed value, or an incorrect mapping, open an issue.

Requirements

• Python 3.10 or newer
• Tkinter available in your Python install
• Windows is the primary supported workflow for automatic tool download

BackBeat requires these external tools:

• yt-dlp
• ffmpeg
• ffprobe

If those tools are not found on PATH or in ./bin, BackBeat offers to download them automatically into the local bin/ folder on Windows. No admin rights are required.

On some fresh Windows systems, yt-dlp may also benefit from an optional JavaScript runtime. BackBeat can offer to install Deno into ./bin for that case.

Quick Start

1. Install Python if needed. Download it from https://www.python.org/downloads/ and make sure Add python.exe to PATH is enabled during setup.
2. Put your CSV file in the same folder as backbeat.py, or use the included backbeat.csv.
3. Run:

python backbeat.py

4. If required tools are missing, allow BackBeat to download them.
5. In the setup window, choose your encoding settings and CSV selection.
6. Click Start.

BackBeat writes output into subfolders named after the CSV Source value, such as RB1/, RB2/, or Unknown_Source/.

Setup Window

Current versions use a single combined startup window called BackBeat Setup.

The left side contains encoding settings:

• Browser cookies
• Quality
• Output format
• WebM encode profile

The right side contains CSV selection controls:

• CSV file
• Source
• Ignore save file
• Mark all as processed
• Choose songs...
• a live row count showing X row(s) in source, Y to process

Choose songs... now opens a sortable table (Treeview) built from the current CSV and Source selection. Only songs with valid URLs are shown.

The picker includes these columns:

• Selected (checkbox marker)
• Status
• Filename
• Notes
• Link (click to open the Youtube URL)
• Source
• Delay
• Speed
• Crop Bars

Status values:

• new: song is not in backbeat_processed.csv
• update available: same Source + Filename exists in cache, but one or more settings changed
• already processed: full cache match with identical settings

Default checkbox behavior:

• new and update available start checked
• already processed starts unchecked

When a song is update available, changed values are wrapped as *value* in the affected columns (Link, Delay, Speed, Crop Bars).

Notes is displayed for convenience in the picker, but it is not used when determining whether a row matches the processed cache.

Picker controls include:

• Select all
• Select unprocessed
• Clear all
• Show filter radios: All, New, Updates, Processed

Sorting:

• click any column header to sort
• click the same header again to reverse
• Selected and Status can be sorted independently

Filtering and sorting do not alter your selection state; they only change what is currently shown and order.

Hover tooltips are available on the labeled controls and checkboxes.

Encoding Settings

Setting	Description
Browser cookies	Lets yt-dlp borrow cookies from your browser session. Useful for age-restricted videos, members-only videos, and Premium-only higher bitrate streams. On Windows, BackBeat tries to preselect your default browser.
Quality	Best available, 1080p max, 720p max, 480p max, or Smallest file.
Output format	webm or mp4. webm is the default and generally the preferred option for rhythm-game video backgrounds.
WebM encode profile	Applies only to webm. Auto selects a profile by source height. Manual options are Fast / Small, Medium / Medium, and Slow / Big.

WebM Profiles

Profile	CRF	Target Bitrate	Max Bitrate	CPU Used
Fast / Small	10	4M	6M	4
Medium / Medium	6	6M	9M	2
Slow / Big	4	8M	12M	0

When Auto is selected, BackBeat chooses:

• Fast / Small for sources up to 480p
• Medium / Medium for sources up to 1080p
• Slow / Big for sources above 1080p

CSV Selection And Filtering

BackBeat lists every .csv file in the script folder except backbeat_processed.csv. If backbeat.csv exists, it is listed first.

Source filtering is case-insensitive when choosing what to process. All processes every matching row in the selected CSV.

Rows with missing or invalid Youtube URLs are ignored entirely. They do not count toward the live row total and they are never processed.

CSV Format

Use this header layout:

Source,Filename,Youtube,Delay,Speed,Remove Black Bar
RB1,my_video,https://www.youtube.com/watch?v=...,0,100,yes
RB2,another_song,https://youtu.be/...,500,98.5,no

Column	Description
Source	Group or setlist name used for filtering and output folders.
Filename	Output file name without extension. Invalid filename characters are stripped automatically.
Youtube	Full http or https URL to the source video.
Delay	Milliseconds to adjust the start. Positive values trim from the start. Negative values add black video at the front. Blank is treated as 0.
Speed	Playback speed percentage. 100 means unchanged. Blank is treated as 100.
Remove Black Bar	Enables crop detection when set to yes, true, or 1.

Processing Cache

BackBeat stores prior successful rows in backbeat_processed.csv.

This cache is based on the CSV input row, not on the chosen startup settings. A row is considered the same if these CSV fields match:

• Source
• Filename
• Youtube
• Remove Black Bar
• Delay with blank treated as 0
• Speed with blank treated as 100

That means changing format, quality, browser cookies, or WebM profile does not invalidate the cache by itself. If you want to re-encode the same rows with new startup settings, use Ignore save file.

Ignore Save File

If checked, BackBeat processes all selected rows even if they already exist in backbeat_processed.csv.

When the manual song picker is active, selected songs already run regardless of cache, so this option mainly affects non-manual runs.

Mark All As Processed

If checked, BackBeat adds all selected rows to backbeat_processed.csv without downloading or encoding anything, then exits.

With manual song selection active, this marks only the rows checked in Choose songs....

Use this only when those rows were already handled elsewhere and you want future runs to skip them.

Output Behavior

• output files are written to a folder named after Source
• if Source is empty, output goes to Unknown_Source/
• the output extension matches your chosen format: webm or mp4
• audio is removed from the output
• if no processing is needed for a row, BackBeat may simply rename the downloaded file into place

At the end of the run, BackBeat prints a summary showing which rows succeeded, failed, or were skipped as already processed.

CON File Compatibility And Nautilus Naming

The generated filenames are intended to line up with CON-format song sets that were batch-renamed with Nautilus using the The Artist - Song style shown below:

When using Nautilus, make sure Ignore Xbox file system limitations is enabled so long or Unicode-safe names are not truncated unexpectedly.

Third-Party Tools

BackBeat can download these tools into ./bin/ when needed:

Tool	License	Source
yt-dlp	The Unlicense	https://github.com/yt-dlp/yt-dlp
ffmpeg	GPL v2+	https://github.com/BtbN/FFmpeg-Builds
Deno	MIT	https://github.com/denoland/deno
Nautilus	MIT	https://github.com/trojannemo/Nautilus

BackBeat itself is not affiliated with YARG, Rock Band, Nautilus, or any other rhythm-game project.

Project Note

This project was built with heavy AI assistance. Most of the saved effort went into curating and syncing the included CSV dataset, which is the time-consuming part of a project like this.