Audio description voice for videos
Audio description tells blind and low-vision viewers what is happening on screen: who enters, what changes, what text appears, spoken in the pauses between dialogue. Web accessibility guidelines ask for it on prerecorded video at level AA (WCAG success criterion 1.2.5), yet recording a narrator for every video and every update is slow and costly.
SuperClonica voices your description script with a clear, neutral voice, free to test online and offline on your own computer. This guide shows how to write descriptions, how to voice them and how to fit them into the video.
What makes good audio description
Describe what is seen, not what it means; let viewers draw their own conclusions.
- Short, present-tense sentences: “She opens the door. The room is empty.”
- Only what matters for understanding: actions, changes, on-screen text, who is speaking.
- Spoken in the natural pauses, never over important dialogue.
- A calm, neutral voice that stays the same across a whole series.
Hear a line free online
On the page of any voice, paste a description line of 30 to 100 characters, choose the language and press Voice it. In a few seconds you hear whether the voice is clear enough for description, with no account.
Write descriptions for the gaps
If a gap is too short, cut the sentence rather than speeding the voice up too much.
- Watch the video and note every gap in dialogue with its start time and length.
- Write one description per gap, short enough to fit: roughly two to three words per second.
- Name each file by its start time, such as ad_0012 for the cue at 0:12.
- Read the text of on-screen titles and captions out where viewers need them.
Voice and mix them offline
Install the free SuperClonica engine: python -m pip install superclonica-cpu. Voice each description into its own file, for example superclonica --text-file ad_0012.txt --voice F1 --lang en --normalize --output ad_0012.wav. --normalize keeps every line at the same loudness, and --speed slightly above the default helps a line fit a short gap. Place each file at its start time in your video editor, lower the programme sound a little under it, and export a described version or a separate description track. The same voice can describe the video in up to 31 languages.
Keep people in the loop
Use a neutral stock voice, F1 to F5 or M1 to M5, or a clear narrator from the catalog, and keep it for the whole series. Whenever possible, ask blind and low-vision viewers to review the result: they know best what is missing. A synthetic voice makes description faster and easier to update, but it does not replace a careful script.