Voiceover
When you click Approve & build, VidNexus makes the narration with ElevenLabs: one audio file per scene. Each clip is saved and reused when nothing changed, so you don't pay twice for the same scene.
One MP3 per scene
Every scene gets its own voiceover file, not one long track for the whole script. Motion graphics, b-roll and captions all follow scene boundaries. With one file per scene, you can redo one scene without paying for the rest again.
What this enables
Redo one scene. Edit the narration in scene 4 and click Regen voice. Scenes 1 to 3 and 5 onward stay as they are. No wasted characters.
Exact scene timing. VidNexus knows exactly where each scene starts and ends, because each one has its own file.
Caching
Each voice clip is saved and reused when the text, voice and voice model are all the same. Then there is no ElevenLabs request and no characters are used. If any of them change, a new clip is made and saved.
When clips are reused
Same script, second run: every clip is reused. The voice step finishes in seconds.
Edited scene 3's text: only scene 3 is made again. The other scenes are reused.
Changed the voice: every scene is made again.
Rebuilt an old project: if the audio files are still on disk, it's still free. They are in the voice folder inside your working files folder (by default
%APPDATA%\VidNexus\assets\voice).
Picking a voice
The Voice tab in the editor shows every voice on your ElevenLabs account: the ready-made voices, any you've cloned, and voices you've added from the ElevenLabs library. Each row has a Preview button that plays a sample.
Voice models
Multilingual v2
The default. Best at pauses and emphasis.
Turbo v2.5
Faster and cheaper per character. A little less expressive than Multilingual v2.
Monolingual v1
Older. Sounds flatter.
The model is separate from the voice: the same voice works with every model. Change the model in Settings > API keys > ElevenLabs model. Leave it empty for Multilingual v2.
Per-channel defaults
Set a voice for each channel, so every new video on it starts with the right narrator. Channels > Edit on the channel > Narrator voice > Save.
When a scene needs a voice, VidNexus uses the first one that is set:
The voice in Video settings (this video only)
The brand kit's default voice
The channel's narrator voice
Default voice ID in Settings > API keys
Per-video overrides
Sometimes one video needs a different narrator, like a guest series or a calmer deep dive. Open Video settings before you click Approve & build, and change the voice for that video only. The brand kit and channel stay as they are.
Common per-video changes
Shorts voice: a livelier narrator for Shorts than for long videos.
Hook A/B: use the Hook A/B generator to hear 8 different opening lines in this video's voice, without touching the rest of the video.
Voice cloning
Voice cloning happens on ElevenLabs. Upload 30 seconds to 3 minutes of clean audio at elevenlabs.io/voice-lab. Once it's made, the voice shows up in the voice list in VidNexus.
Cloning source-audio tips
Single speaker, no background music or noise.
Natural speech, not read-aloud. Interview clips work well.
Match the tone you want. For documentaries, use a calm, documentary-style recording.
44.1 kHz or higher, mono, uncompressed if possible.
Voice settings
Besides the model and voice, ElevenLabs has four voice settings. VidNexus always uses the values below. You can't change them in the app. Speech speed stays at natural (1.0): faster settings made almost no difference you can hear.
Voice parameters
Stability (0.4)
Lower sounds more emotional but varies more. Higher sounds flatter but steadier.
Similarity (0.75)
How closely the voice sticks to the original.
Style (0)
Exaggerates the voice's own way of speaking. Off.
Speaker boost (on)
Makes the voice sound a bit more like the original. Helps most on cloned voices.
Delivery pace
Voices speak at different speeds with the same settings, so the voice you pick sets the pace. Click Listen in Video settings to compare a few. To plan a video's length, open Tools > Script length and pick a delivery pace (Fast, Balanced or Slow).
To add pauses, use commas or ellipses in the narration. ElevenLabs reads punctuation as pauses. To speed it up, use shorter sentences and contractions.
What's next
Stock footage: How the b-roll under your narration gets picked.
Render pipeline: How voice, footage, motion graphics and captions become the MP4.