New music models: Google Lyria 3.5, Eleven Music 2.5 and Eleven v4

Hi there,

The last two emails were about picture: shots planned as a sequence, then rendered with everything the scene knows. This one is about sound. Music in Frameliq is now written for the scene, not for the shot, so it no longer restarts at every cut.

1. Two music models to choose from

The AI Music section in Composer's Audio tab now has a model picker.

• Eleven Music 2.5 gives you the exact length you ask for. It costs about 60 Media credits per started minute, down from 80.
• Google Lyria 3.5 costs about 10 Media credits per track, whatever the length. It decides the length itself, so Frameliq trims the track to fit.

Your choice is remembered as you move between shots.

2. One cue for the whole scene

Open any shot in a scene that has several shots and the music is set to "Whole scene" by default. The cue is as long as the scene, it starts under the scene's first shot, and every shot of that scene plays under it. Generate again and the new cue replaces the old one. You can still switch to "This shot" for a single sting.

3. A score direction, and a prompt written for you

The tab now shows a short score direction for the scene. It comes from your genre, where the beat sits in the story, how tense it is, and whether anyone speaks. Under dialogue it asks for a sparse cue with no lead melody, and the export still lowers music under spoken lines.

Press Write and Frameliq turns that direction into a prompt for the model you picked. If you would rather not write one, leave the prompt empty and pick a mood and a tempo.

4. Shots no longer bring their own background music

Video models that make sound used to add music to each shot, and it changed at every cut. Shot prompts written by Frameliq now ask for dialogue, ambience and effects only, and the score comes from the cue on the timeline. Music that is played inside the scene, like a radio or a band, is still written into the shot.

5. Voices move to Eleven v4

Voice previews, voice masters, and dialogue spoken by one voice now use ElevenLabs' newer Eleven v4 model. It also costs less: about 8 Media credits per 1,000 characters instead of 10. Takes where several characters speak together stay on Eleven v3 for now. There is nothing to change in your project.

Smaller things you'll notice

• When you send storyboard frames to Composer, each frame has a checkbox, with a count and Select all, so it is clear what will be imported.
• Change a shot's model and Composer offers to rebuild the prompt for the new model.
• Stories written by Frameliq, and by Claude or ChatGPT through the Frameliq connection, now give every beat a music prompt that keeps one key, tempo and set of instruments across a scene.
• The dialogue button and voice previews show the Eleven v4 price before you generate.
• "Generate shot" now lives on the Compose tab, so it no longer covers the Audio and References tabs. It stays in view while a shot is rendering, so you can always stop it, and a lip-sync shot keeps it on every tab.
• "Add sound to this shot" and "Add sound effects to this shot" now take you straight to Compose to generate.
• If a save is refused or your connection drops, Frameliq now tries again on its own instead of waiting for your next edit.

No fal key? You don't need one

Most people who stall in Frameliq stall at the same step: images, video and sound ask for a fal key, and they don't have one. There are two ways past it:

• Creator Plus, $29/mo: 800 Media credits a month are included, so stills, video, voices and music work with no fal account.
• Creator, $19/mo: add your own fal key in Settings and pay fal directly for what you generate.

Hit reply and tell us how your film sounds now. We read every reply.

Happy filmmaking,

The Frameliq Team