Configuration Reference

All channel behavior is controlled by three JSON configuration objects stored on each channel record. No code changes are needed to customize a channel — everything is database-driven.

What It Does

Configuration determines how every pipeline stage behaves for a given channel: which voice to use, how the video looks, whether to include background music, what subtitle style to apply, and more. Changing these settings immediately affects the next pipeline run without any redeployment.

voice_config

Controls text-to-speech generation.

Field	Type	Default	Description
`voice`	string	`"alloy"`	OpenAI TTS voice: alloy, echo, fable, nova, onyx, shimmer
`speed`	number	`1.0`	Speaking speed multiplier (0.25 to 4.0)

visual_config

Controls video composition, backgrounds, subtitles, and thumbnails.

Background Settings

Field	Type	Default	Description
`background_color`	string	`"#1a1a2e"`	Hex color for solid background fallback
`background_image`	string	`null`	Storage path to a template background PNG
`overlay_image`	string	`null`	Storage path to a transparent overlay PNG

Slideshow Settings

Field	Type	Default	Description
`slideshow_provider`	string	`null`	Image provider: `"pexels"` or `null` (disabled)
`slideshow_image_count`	number	`5`	Number of images to fetch per video
`slideshow_keywords_source`	string	`"auto"`	`"auto"` (extract from script) or `"manual"`
`slideshow_manual_keywords`	array	`[]`	Manual keyword list (if source is manual)
`slideshow_duration_per_image`	number	auto	Seconds per image (auto-calculated if null)
`slideshow_zoom_range`	array	`[1.0, 1.3]`	Ken Burns zoom min/max multiplier

Subtitle Settings

Field	Type	Default	Description
`subtitle_color`	string	`"#ffffff"`	Main subtitle text color (hex)
`subtitle_highlight`	string	`"#e94560"`	Accent/highlight color for the currently spoken word
`subtitle_style`	string	`"word_highlight"`	Subtitle mode: `"word_highlight"` or `"simple"`
`font`	string	`"Montserrat-Bold"`	Font family for subtitles

Branding

Field	Type	Default	Description
`accent_color`	string	`"#e94560"`	Brand accent color (thumbnail bar, UI highlights)

music_config

Controls optional background music mixed under the voiceover.

Field	Type	Default	Description
`provider`	string	`null`	Music provider: `"jamendo"` or `null` (disabled)
`volume`	number	`0.15`	Music volume relative to voice (0.0 to 1.0)
`fade_in_seconds`	number	`1.0`	Music fade-in duration at start
`fade_out_seconds`	number	`2.0`	Music fade-out duration at end

Other Channel Fields

These are top-level fields on the channel record, not nested inside a config object:

Field	Type	Description
`name`	string	Display name (e.g., “Gaming”)
`slug`	string	URL-safe identifier (e.g., “gaming”)
`prompt_template`	string	Channel-level context injected into all AI prompts
`posting_cadence`	number	Target videos per day (controls topic selection count)
`active`	boolean	Whether the channel is active in the pipeline

Where to Find It

Dashboard: Channel Settings page — each config section has its own tab (Voice, Visual, Music)
Database: channels table — voice_config, visual_config, and music_config columns

How It Works

When any pipeline stage runs, it reads the channel’s current configuration from the database. This means:

You change a setting in the dashboard (e.g., switch voice from “alloy” to “nova”)
The dashboard writes the updated voice_config JSON to the channels table
The next pipeline run reads the new config and uses it

There is no caching or delay. Changes take effect on the very next pipeline execution.