Skip to main content

How to create a voice-over for a video in Free-Flow

A voice-over replaces the speech of a video with an AI voice that reads your subtitles, sentence by sentence, at the moment each one was spoken. This article shows how to generate a voice-over in Free-Flow, check that it fits the video, and render the final video.

Before you start

You need a Free-Flow document with a video and subtitles in the language of the voice-over. See How to subtitle a video in Free-Flow to upload, transcribe and translate a video. Free-Flow is part of the wxrks Translation Portal, so the portal must be turned on for your account.

  • Account Admins turn the portal on in Settings > Portal & Routing > Portal and choose, on its Access tab, which roles land on the portal after sign-in. They create voice-overs from Tools > Free-Flow.

  • Project Managers create voice-overs from Tools > Free-Flow, with no setting to change.

  • Clients create voice-overs from Tools > Free-Flow, or open Free-Flow from the Translation Portal sidebar.

The voice-over reads the variant on screen. To voice a translation, switch to that variant in the top bar first.

How the voice-over fits the video

Free-Flow keeps the voice in time with the picture:

  • The script is split into sentences, and each sentence starts when it was spoken in the original video.

  • Each sentence has a slot that ends just before the next sentence starts.

  • A sentence that is too long for its slot is sped up slightly. If it still does not fit, the following sentences start a little later.

Translations are often longer than the original; Portuguese, for example, usually takes more time to say than English. Free-Flow reports every sentence that does not fit, so you can fix it before rendering.

Generate the voice-over

  1. Open the document, switch to the variant you want to voice, and open the Voice-over tab in the side panel.

  2. Choose a Voice. The list shows voices for the language of the variant and multilingual voices; Filter voices narrows it. Click the play button next to the list to hear a sample.

  3. Optionally set the Speed (Pace for some voices) and the Style of the voice.

  4. Click Generate voice-over.

Voice-over tab with badge 1 on the Voice list, badge 2 on Speed and Style, and badge 3 on Generate voice-over.

The new voice-over appears in the Voice-overs list, where you can download its audio or delete it. Free-Flow also separates the background of the original audio (music and effects) so it can be kept under the new voice; Background track ready shows when it is done.

ℹ️ Note: If you edit the subtitles after generating, the voice-over is marked Out of date. Generate it again: sentences that did not change are reused, and only the others are synthesized again.

Check the fit and preview

  1. Turn on Play with video and play the video to hear the voice-over with the picture. Use the Original, Background and Voice sliders to balance the preview, and Hear the original to compare.

  2. Open the fit report below Edit audio. It says Every segment fits its slot, or lists each sentence that does not fit with how much it is off, for example 0.4s too long or 1.0s late.

Voice-over tab after generating: badge 1 on Play with video and badge 2 on the fit report listing four segments that do not fit their slot.

To fix a sentence that does not fit:

  • Shorten the text of the cue in the subtitles, then generate the voice-over again.

  • Choose a faster Speed and generate again.

  • Adjust the audio by hand with Edit audio (see below).

Render the video

  1. Under Subtitles, choose how subtitles appear in the video:

    • Burned into the picture: always visible. The picture is re-encoded, so rendering takes longer.

    • Subtitle track: shown only when the viewer turns subtitles on in the player.

    • No subtitles.

  2. Turn on Keep the original audio to add the original speech as a second audio track, off by default, that viewers can switch to from the player's audio menu.

  3. Click Render video.

When the render is ready, it appears in Rendered videos. Click Download video to save the MP4.

Render options with badge 1 on Subtitles, badge 2 on Keep the original audio, and badge 3 on Render video.

Edit the audio by hand

Edit audio opens the voice lane under the timeline, with one clip per sentence. Changes are saved as a new version of the voice-over, so the original stays available.

  • Drag a clip to move it, and drag its edges to trim it.

  • Hold Alt (⌥ on macOS) while dragging an edge to speed the clip up or slow it down.

  • Press S to split the clip at the playhead and Backspace to delete it.

  • Select a clip to change its Clip volume (dB), Fade in (ms) and Fade out (ms), or click Fit to cue to fit it to its subtitle.

When you are done, enter an optional Version name and click Save as v2 (the number follows the last version). Click Discard to leave without saving.

Edit audio mode: badge 1 on the voice lane with one clip per sentence under the timeline, and badge 2 on the editing panel with Version name, Discard and Save as v2.

Videos with several speakers

If the subtitles mark speakers (turn on Speakers before transcribing, or assign speakers to cues), the Voice-over tab lists them under Speakers. Each speaker uses the voice-over's settings by default; open a speaker's settings to give it its own voice. Turn on Let speakers overlap to let a line start up to half a second before another speaker's line ends.

Quick reference

Item

Value

Subtitle options

Burned into the picture, Subtitle track, No subtitles

Original audio

Optional second audio track

Output

MP4 video; the voice-over audio can also be downloaded on its own

Voice-over versions

Each Edit audio save creates a new version

Did this answer your question?