Voice cloning

Voice cloning creates a synthetic replica of any voice from a short audio sample and saves it as a private workspace voice. Once created, the cloned voice is available for use in Text to Speech and Nabrah Agents.

Guide

1. Provide audio

Two input methods are available:

Record record directly in the browser. Recording stops automatically at 20 seconds. A sample Arabic sentence is provided on screen to read aloud if needed.

Upload drop a WAV file or click to browse. Maximum file size is 4 MB and maximum duration is 20 seconds.

2. Preview and trim

Once audio is added, the waveform appears in the preview panel on the right. Trim a 4–7 second segment from the recording to use as the clone source. Aim for a segment with clear, natural speech and no silence or background noise at the edges.

3. Name and clone

Enter a voice name, then click Clone voice. The cloned voice is saved to the workspace and available immediately under Voices and in the Text to Speech voice selector.

Best results

• Record or upload up to 20 seconds of clear speech.

• Avoid long silences and background noise.

• Match the pacing and volume intended for the final output the clone reflects the characteristics of the source recording.