Enable the AI Lyric Generator

If you want the app to help you write lyrics and a ready-to-paste Suno style prompt before you render — instead of bringing your own MP3 — turn on the AI Lyric Generator. Everything runs locally on a language model you download once; nothing about your song ideas is sent to the cloud.

This page is a setup checklist. For the full feature reference, see AI Lyric Generator.

Before you start

Make sure you’ve already done these:

• Set up your brand — the voice description you write there is the backbone of the generator.
• Checked your RAM. The smallest model tier needs 8 GB minimum (12 GB recommended); the biggest tier wants 24–32 GB. Older laptops will struggle on anything above the Light tier.
• Cleared a few GB of disk space. Models range from ~5 GB to ~13 GB.

1. Turn the feature on

1. Open Settings → Preferences.
2. Under Enable features, toggle AI Lyric Generator on.

A new Lyrics tab appears in Settings. The Create flow also gains two extra steps — Lyrics and Song — which you’ll start using in the next doc.

2. Pick a model tier

Open Settings → Lyrics. You’ll see four model cards. The app detects your system RAM and suggests a default; models it doesn’t think your machine can run well are flagged but still selectable.

Tier	Model	Size	Min RAM	Recommended	Best for
Light	Qwen 2.5 7B Instruct	~5 GB	8 GB	12 GB	Low-spec laptops, older Macs
Balanced	Qwen 2.5 14B Instruct	~9 GB	16 GB	24 GB	The default for most machines
Best	Mistral Small Instruct 2409	~13 GB	24 GB	32 GB	24+ GB Apple Silicon, high-VRAM PCs
Uncensored	Cydonia 22B v2q	~13 GB	24 GB	32 GB	Same as Best, when you want direct language on explicit topics

If in doubt, start on the tier the app recommends. You can download additional tiers later and switch between them.

3. Download the model

1. Click Download on the tier you want.
2. A progress bar appears on the card. Downloads stream directly from HuggingFace to <app data>/models/llm/.
3. Click Cancel any time to abort — partial files are cleaned up.
4. When the download finishes, the card gains an Installed badge and a Delete button.

You can install multiple tiers side-by-side and switch between them later.

4. Finish your brand

Scroll down on the same Settings → Lyrics page to the Your Brand form. The voice you wrote in Set up your brand should already be here — fill in the rest:

Required:

• Genre — pick a preset lane or choose Custom… to write your own short genre description.
• Vocal type — female, male, or duet.
• Explicit content — Off keeps lyrics non-explicit, On allows direct language when topics call for it, Contextual lets each song’s topic decide.

Optional but worth it:

• Themes you gravitate toward — emotional territories the brand rotates through.
• Never use — words, images, or tropes the generator should avoid.
• Always prefer — detail types you want leaned on (proper nouns, specific times, body-level detail, etc.).
• Artist references — artists or songs you want to sound adjacent to. Used as inspiration, not imitation.

Fields auto-save as you type.

5. You’re ready

Head to the Create tab — you’ll land on the new Lyrics step instead of the drop zone. Keep going with Make your first video with AI lyrics.

Turning it off

Toggle AI Lyric Generator off in Settings → Preferences and the Create flow reverts to the plain drop-an-MP3 experience. Your installed models and brand data are kept — you can turn the feature back on without re-downloading or re-configuring anything.