RTC-SDD: Speech Deepfake Detection Challenge
An ICASSP 2027 Grand Challenge to detect deepfake speech after it has passed through real-time communication platforms.
About RTC-SDD: Speech Deepfake Detection Challenge
There are three tracks: black-box RTC processing, robustness to RTC distortions, and generalisation to unseen speakers, generators and platforms. Final evaluation starts 9 November 2026, final submissions close 16 November 2026 (AoE) and results come out 23 November 2026. The top five teams are invited to submit a two-page ICASSP 2027 paper (due 7 January 2027). The page lists no cash prizes.
Who can apply
Individuals or teams; each participant may be on only one team; members need not share an affiliation; registration information in English. No registration deadline.
How to get it
- Create a Codabench account
- Fill in the team registration form and apply on Codabench
- Request dataset access on Hugging Face and submit before 16 Nov
RTC-SDD: Speech Deepfake Detection Challenge — FAQ
Who can apply for RTC-SDD: Speech Deepfake Detection Challenge?+
Individuals or teams; each participant may be on only one team; members need not share an affiliation; registration information in English. No registration deadline.
How do I get RTC-SDD: Speech Deepfake Detection Challenge?+
1. Create a Codabench account 2. Fill in the team registration form and apply on Codabench 3. Request dataset access on Hugging Face and submit before 16 Nov
How much is RTC-SDD: Speech Deepfake Detection Challenge worth?+
Research challenge. An ICASSP 2027 Grand Challenge to detect deepfake speech after it has passed through real-time communication platforms.