ANITA addresses one of the strongest needs of PMR (Professional Mobile Radio communication) users: to significantly improve the quality of their communications in particularly harsh acoustical environments. Indeed, speech clarity is critical for many PMR users as fire brigades or police forces. Siren noise, alarms, car engines can result in a real threat to the workers' and citizens' safety. ANITA proposes to develop innovative DSP-based solutions for in-car and street applications. Environmental noises will be removed in quasi real time (<25 ms) to facilitate communication (Mean-Opinion Score >=3, SNR before transmission >= 10), significantly improving users' working conditions and work efficiency. Technical approach include innovative noise reduction algorithms combined with microphone arrays. Algorithms will be implemented on a PC platform for real-time demonstration and into DSPs for prototyping.
A consortium of 5 partners has been constituted to gather their complementary expertise and ressources for the ANITA project.
- EADS Defence & Security Networks (F);
- coordinator: specialised in secured telecom networks - technical specifications, integration & validation;
- FULCRUM VOICE TECHNOLOGIES (UK): advanced products related to voice technology - algorithms development & integration;
- KTH (S): research centre in speech processing - algorithms development;
- FAU (D): research centre in microphone arrays - algorithms development
- TRADIA (E): telecom network operator - users requirements, in-site validation.
The ANITA project relies on the development of innovative algorithms developed for noise cancellation and speech enhancement. The technical approach combines microphone arrays (advanced beamforming algorithms to focus on the speaker signal, acoustic echo cancellation when a reference signal is available (e.g. sirens), blind source separation algorithms for suppressing coherent interferers) with a single-channel signal processing technique using a-priori knowledge of both noise and speech signals.
The work programme has been divided into 7 workpackages, for a 30-month duration:
WP1: Project management;
WP2: End-users requirements, technical specifications and voice recordings;
WP3: Microphone array signal processing;
WP4: Single-channel approach;
WP6: In-site validation;
WP7: Dissemination & exploitation.
Algorithms will be first implemented on a PC platform for real-time demonstration and integrated into DSPs in the second half of the project.
- an external noise reduction module for in-care application, applied to current PMR systems (TETRA-TETRAPOL-APCO25);
- a handset (embedded solution) for pedestrian mode built for next-generation PMR.
A strong strategic impact is expected in the PMR market led by the US. The main milestone has been set at month 18. The consortium will assess the computational complexity of the algorithms and the amount of manpower to allocate for their integration into DSPs.
Funding SchemeCSC - Cost-sharing contracts
08908 Hospitalet De Llobregat
HP12 4HJ Buckinghamshire
100 44 Stockholm