Cross-Lingual Voice Cloning (IWSLT 2026 Shared Task)

[Yasmin Moslem]

Dear Colleagues,

If you are interested in speech synthesis and translation, we are organizing the Cross-Lingual Voice Cloning Track at IWSLT 2026. This shared task addresses one of the most challenging areas in speech technology, cloning a speaker's voice across different languages.

The Challenge: Given a small set of audio samples from a speaker in a source language, the task is to synthesize natural-sounding speech for unseen content in a target language, while preserving the speaker's voice characteristics.

Why does this matter? Voice cloning is beneficial in multiple domains:

• Speaker Standardization: Streamlining voices in speech systems efficiently without recording everything from scratch
• Education: Creating age-appropriate, culturally relevant voice experiences
• Accessibility: Personalizing assistive communication tools

Important dates:

1st April 2026	Release of blind test data
15th April 2026	System submission deadline
24th April 2026	System paper submission deadline

Task details: https://iwslt.org/2026/voice-cloning

If you have any questions, please feel free to contact me. Looking forward to your participation!

Best regards,

Yasmin Moslem