Voice and visuals
Voice and visuals: local narration, stock clips and AI only if you enable it
By default, the voice and visuals of every CLAQON video come from your own machine and from licensed stock: a local narration with each word’s timing for the captions, and vertical clips from Pexels. ElevenLabs voice, AI images and AI video come switched off; you enable them per account, with their budget and their cost in plain view.
- Local narration with no per-video cost
- Licensed stock clips by default
- Paid options off until you enable them
What it solves
How do you narrate and illustrate every day without sending costs soaring?
Voice and, above all, AI-generated video are the expensive part of a short. If every piece pays for everything, the cost is multiplied by the number of accounts and slots; if nothing is paid for, the result can fall short in the scenes that matter.
CLAQON starts at zero – local voice and stock – and lets you pay only where it adds something: a paid voice, generated images or AI video clips in the key scenes. Each paid option is enabled per account, with its own budget and with the estimated cost shown before spending.
How it works
From narration to the edit
Narration
Piper or Kokoro on your machine, ElevenLabs if you enable it, or your own aligned voice. Always with each word’s timing.
Scene plan
The real timing of the narration decides how long each scene lasts.
Visuals according to the profile
Stock by default; with a paid profile, AI images or video in the key scenes.
A gate before paying for video
No AI video clip is requested until 7 conditions are met, from the approved script to the references for every scene.
An edit with sound
FFmpeg edits on your machine, with music chosen by tone and effects that rotate.
Capabilities
Voices, visuals, music and effects
Voice for your videos
- Local voice with Piper
- Narration on your machine, with no per-video cost and word-level timing. Some voices come preinstalled and the rest are downloaded when you use them.
- Kokoro in English
- A more natural local voice for US and UK English, with 6 voices.
- Licences of the local voices
- The local voices have different licences and not all of them allow commercial use. The commercial edition will be prepared with an allowlist of voices with a commercial licence: until then, check each voice’s licence before using it on a monetised channel.
- ElevenLabs Off by default
- Multilingual v2, Flash v2.5 and v3, with word-level timing. It uses the voices in your account, asks you to confirm the right to use them and validates voice and model before spending.
- Your own narration
- You upload your audio (25 MB at most) and Faster Whisper aligns it locally with the script. If it does not match, it is rejected without paying for another synthesis.
- Account language
- The narration uses the account’s language, and you get an alert if the script is written in another one.
- Pronunciation dictionary Beta
- Per account. Today it applies to local narration in the Studio and in the Commentary Room; not yet in the factory.
Visuals: stock, AI images and AI video
- Contextual stock
- Licensed vertical clips from Pexels, scored per scene and without repeating anything the account used in the last 45 days. It is what every account uses by default.
- Visual profiles
- Free, Saver, Balanced, Quality and Strict decide how much visual AI is paid for, from stock only to generated clips in every scene. The panel shows the estimated cost of each profile before you apply it.
- Key shots Off by default
- In hybrid mode, a deterministic selector picks 1 to 8 key scenes (hook, climax, ending…) for an AI video clip; the rest use stock or images.
- AI images Off by default
- FLUX.2 from BFL, in its klein 4B, klein 9B and pro variants, up to 1 megapixel.
- AI video, from image to video Off by default
- 18 general models: 12 through Runway (such as Gen-4 Turbo, Veo 3.1 or Wan 3) and 6 through fal.ai (H3 Max and PixVerse V6). Each clip starts from a validated reference image.
- Regenerate a scene
- Up to the maximum you configure. The voice and the rest of the assets are kept.
- Encrypted connections
- Each provider’s keys are stored encrypted with AES-256-GCM and are never sent back to the browser. You can see the plan, the quota or the balance at no cost, depending on what each provider allows.
Music and sound effects
- Music by tone
- 9 music beds under a CC0 licence (curiosity, tension, energy and calm), levelled to −20 LUFS and chosen by the tone of the script. You can also upload your own music.
- Sound effects
- 18 effects: whoosh, impact, riser and pop, plus comic, ding and glitch for the AI director. The variants rotate so they do not repeat, and each one carries its provenance in the video’s record.
- Loudness and noise Off by default
- Normalisation to −14 LUFS for social platforms and noise reduction per edit, in the Studio and in the Subtitle Room.
Limits
Limits of voice and visuals
- Not all the material shares the same licence: not all local voices allow commercial use, we do not claim that all the effects allow it, and Freesound and GIPHY, which the AI director uses (off by default), have their own terms. Check the provenance record before monetising.
- No voice cloning, no dubbing and no regenerating a single sentence of the narration.
- Stock comes only from Pexels: there are no large creative libraries built in.
- AI images and video need your BFL, Runway or fal.ai keys and budget in the account. ElevenLabs requires a paid plan for commercial use.
- The price catalogue expires every month: if it expires, nothing paid is quoted, not even the ElevenLabs voice.
Frequently asked questions
Questions about voice and visuals
Can I produce without paying for voice or images?
Yes. It is the default configuration: local narration on your machine, stock clips from Pexels and editing with FFmpeg. Apart from the free Pexels key, you only need an AI model for the script, which can be a local one.
Can I use the local voices on a monetised channel?
Not all of them: they have different licences and some do not allow commercial use. The commercial edition will be limited to an allowlist of voices with a commercial licence, which is being prepared. The free ElevenLabs plan does not allow it either.
Can I narrate with my own voice?
Yes. You upload your narration (up to 25 MB) and Faster Whisper aligns it locally with the script so the captions follow every word. In the Editing Studio you can also record yourself with the microphone. CLAQON does not clone voices.
Which AI video models does it support?
Image-to-video models from Runway and fal.ai: 18 in the general catalogue. They come switched off, need your keys and, before paying for a clip, the video gate requires everything before it to be approved.
How much does visual AI cost?
It depends on the profile and the model. The panel shows you the estimated cost of each profile before you apply it, and each production is quoted and reserved before spending, within your caps. See spend control.
Early access
You decide what each video pays for
Tell us what voice and visual style you are looking for on your channels and how much you want to spend per piece. We will tell you how it would be set up in CLAQON.