Sometimes you have no recording to clone — just an idea of how a voice should sound. That's what /voice-design is for: you describe a character (say, "a calm late-night radio hostess, deep voice, slow pace") and the app builds a brand-new synthetic voice from that description. It differs from /voice-clone, which imitates an existing sample; here a voice that never existed before is born. The result is a durable voice_id saved on the canvas that you reuse across your narration. Note: designed voices are experimental and work best in English, Chinese and Japanese.
What you'll need
- An ElevenLabs provider configured on a paid plan (voice design is a paid feature).
- A clear idea of the voice character — age, gender, tone, plus optionally accent and pace.
- An open project in the Workspace, where the preview card and the voice registry card will land.
Step by step
- 1
Launch the /voice-design skill
In the chat panel type /voice-design or pick the skill from the skills hub. Voice design (a brand-new synthetic voice from a description) is different from /voice-clone, which imitates an existing recording — here you need no sample at all.
Film Movie StudioRenderLullaby scoreNew projectSearch skills…AllWorkflowsQuickvideoShort DramaViral vertical mini-series, end to end.imageImage CreatorGenerate single images from a prompt.audioAudio StudioVoiceover, SFX and music tracks.copyApp BuilderMulti-screen apps with live preview.imageE-commerce DesignPro product shots for any platform.videoMusic VideoBeat-synced clips cut to your track. - 2
Describe the voice in your own words
Type into chat how the voice should sound — character, age, gender, tone, accent and pace. E.g. "an energetic male sports commentator in his 30s, loud and fast-paced". The more specific, the better; a vague "a nice voice" won't do.
Film Movie StudioRenderLullaby scoreNew projectSkills/voice-design/short-drama/image-creator/audio-studio/app-builder/voice-design A gentle winter lullaby — soft piano and music-box melody, sweeping strings, dreamlike.
SkillsModelsType / to switch skills anytime
Tip: A proven structure: role + character traits + voice characteristics (age, tone, accent) + delivery pace. - 3
Answer the clarifying question
If the description is too vague, the app asks one question covering the missing gaps (age, tone, accent, pace). Fill them in a single reply. The app then shows the final English voice description it will use to generate.
Film Movie StudioRenderLullaby scoreNew projectFiles
frame_1.png60%promptframe_01.pngGenerating… 42%RemixoutputTip: Chat with the app in your own language, but the voice description itself is always sent in English. - 4
Preview a test line
The app calls design_voice and a preview audio card lands on the canvas automatically. Play it and judge whether the voice fits. If you want to hear a specific line, supply it up front — it must be at least 100 characters, otherwise a preview is generated automatically.
Film Movie StudioRenderLullaby scoreNew projectFilesframe_1.png60%promptframe_01.pngGenerating… 42%Remixoutput - 5
Save the voice on the canvas (voice registry)
The app saves the result as a document card "<voice name> — <voice_id>" with the description, language and best uses; the preview audio is linked beneath it. This card IS the voice's memory — its voice_id lets you reuse the voice in later sessions.
Film Movie StudioRenderLullaby scoreNew projectFilesframe_1.png60%promptframe_01.pngGenerating… 42%RemixoutputTip: Note the voice_id: it's the key you'll use to call the voice any time later. - 6
Generate a longer demo or tweak
The app offers a choice: generate a longer demo, tweak the description and design again, or finish. On "tweak", say what to change (e.g. a slower pace) and the voice is redesigned; the registry card is updated to the new voice_id.
Film Movie StudioRenderLullaby scoreNew projectFilesframe_1.pngframe_01.pngRemixVariationsAdd to timelineFullscreenDelete - 7
Reuse the voice in your narration
In any later generation task, reference the saved voice_id as the voice argument of generate_audio (kind "voiceover"). That way you narrate an entire film with one consistent voice you designed yourself.
Film Movie StudioRenderLullaby scoreNew projectFilesframe_1.png60%promptframe_01.pngGenerating… 42%RemixoutputTip: Have new audio added beneath the voice registry card so the graph reads cleanly: voice card → preview → narration.
What you get
From a plain description you've created a brand-new voice with its own voice_id, saved as a registry card on the canvas alongside its preview audio. You can now reuse this voice across the narration of an entire film — with no recording at all and no dependence on someone else's sample.
Pro tips
- Designed voices are experimental — they sound best in English, Chinese and Japanese; expect less stability in other languages.
- When tweaking, never run the design with an identical description; change at least one detail (pace, tone, accent), or you'll get the same result.
- If design_voice returns a plan or permission error, it's a paid ElevenLabs feature — check your plan; don't retry the call blindly.
Try it yourself
FilmMovie Studio is a desktop app that walks you from idea to finished export. Download it and start creating.
Get FilmMovie Studio