The US Dept of the Interior and IARPA are seeking proposals to develop novel systems that will modify spontaneous speech in real-time to protect and individual’s privacy.
NOTE: Non-US organisations must participate as a collaborator or sub-contractor to a US lead organisation. There may be special requirements in relation to export controls, non-disclosure and security controls.
The ARTS program is seeking innovations to anonymise speech in near real-time by addressing the following three technical areas:
- TA1 – Speaker De-Identification – modify speech signals so that speaker ID systems are unable to recognise the speaker
- TA2 – Modification of Static Straits in Speech – remove existing traits in speech signals related to dialect, gender and age; and replace them with pre-selected traits
- TA3 – Removal of Dynamic Traits in Speech – modify speech signals so that extracted features can’t be used to differentiate short-term states such as emotion and psychological state.
ARTS will be 3 year program with 2 phases of 18 months each (Base & Option phases). Both phases and all technical areas must be addressed by proposals.
Broad Agency Announcement – ARTS: DOI-BAA-ARTS-FY24-01