AI Voiceover

Overview

Sovran's AI voiceover converts written scripts into natural-sounding speech. Every voiceover comes with word-level timestamps so captions are automatically synced.

Generating a Voiceover

1. In your project, select Voiceover in the sidebar, or select Create, then Generate a voiceover. Sovran opens Add a voiceover in Remixes.
2. Choose Generate and enter or paste your script. You can also generate scripts from Brief.
3. Select the voice name to open the list. Preview a voice, then select it.
4. Select the language (15+ supported, including English, Spanish, German, French, and more).
5. Click "Generate".

The voiceover audio is saved to Assets when it is ready. In Remixes, it is also selected for the next batch. Use Upload in the same panel to add your own narration.

Choose a voice and delivery

The selected voices and language stay visible. Select the voice name to open the list. Use All, Google, or ElevenLabs when ElevenLabs is available. Ready custom voices appear with Google voices. Preview a voice before you select it. Your search and filters stay when you open the list again.

Search by name or open Filters. Google voices have gender, pitch, voice role, and use case filters. ElevenLabs voices also have accent filters.

Select Browse ElevenLabs library to find more voices. Search by name, filter by accent, gender, age, or use case, or paste a voice ID or library link. Save up to 10 library voices per workspace. Saved voices are shared with your team. Voices that charge an extra rate cannot be used.

Google preset voices use Gemini 3.8 Flash-Lite TTS. Voice role describes the voice, such as Tutor or Commercial Voiceover. Style controls how it reads your script.

Open Style to choose Natural, Whisper, Friendly, Narration, Promote, or Calm. These presets also work with ElevenLabs. Custom style instructions apply to Google voices. The style applies to the whole read.

For a written script, use Add expression to insert a sound or pause at the cursor. For example: <gasp> That is a great idea! <short pause> Try it today. Expression tags stay out of captions.

Custom voice clones use natural delivery. Gemini styles and expression tags are not available for these voices.

Voiceover generation in this panel uses no render credits. You can request up to 60 reads per person in an hour. ElevenLabs also has a limit of 30 reads per workspace in 24 hours. Shared service limits can make ElevenLabs unavailable sooner. Choose a Google voice or try again later when a limit is reached.

Upload your narration

Open Upload in Add a voiceover. Choose an MP3, M4A, WAV, AAC, or FLAC file up to 64 MB. Use 5 to 120 seconds of audio.

Wait for the transcript. Listen to the audio and correct the transcript before you select Approve and select. This confirms the text used for timed captions. Uploaded narration must be approved before it can be used in a remix.

If you close the panel before approval, return to Upload and check Waiting for review. Open the upload to finish its review, or select Discard.

Clone a voice

Open Manage custom voices, then select Clone a voice. In the dialog, record a sample, upload a file, or select Use Asset Bank video. Use one clear speaker. Sovran isolates the voice before cloning. A clear sample gives the best result.

• Audio uploads accept MP3, M4A, WAV, and WebM. Use 10 seconds to 5 minutes of speech in a file no larger than 20 MB.
• MP4 uploads and Asset Bank videos can be up to 500 MB. Sovran uses up to the first 5 minutes of audio. The sample must contain at least 10 seconds of speech.
• MP4 uploads have their audio extracted in your browser before upload.

Listen to the sample before you continue. Give the voice a name and confirm that it is your voice or that you have permission to use it. Voice creation uses one voice clone credit. The finished voice is available to all workspace members.

Voice isolation can take a few minutes. After the sample uploads, setup continues if you close the dialog.

Delete a custom voice

Open Manage custom voices. Select the delete button beside the voice and confirm. The voice is removed for everyone in the workspace. Any used clone credit is returned. Existing audio stays available.

Word-Level Caption Sync

Every generated voiceover includes precise word-level timestamps. When you use this voiceover in the video editor, captions are automatically placed and timed — no manual synchronization needed.

For an available Sovran AI voiceover with verified word times, use Replace spoken words to change a phrase in the original voice. Preview the result before you apply it. Uploaded narration cannot use this action.

Voiceover Remixes

Use Voiceover Remixes to combine a voiceover with analyzed B-roll, product demos, and reaction clips. Sovran creates editable 9:16 drafts that match the length of the voiceover.

You can select up to three voiceovers and make several drafts for each voiceover. Add captions before you create the drafts, or refine them later in the video editor.

Multi-Language Ads

Generate voiceovers in multiple languages from the same script to test international markets without hiring voice actors or translators.