Updated on 17 July 2026
Advanced Real-Time Speech Enhancement for Edge AI and Human-AI Interaction
CEO / CTO at CandyVoice
Strasbourg, France
About
CandyVoice develops an advanced real-time speech enhancement technology designed for AI-driven voice interaction systems operating under challenging acoustic conditions.
Unlike conventional noise suppression solutions, our technology removes background noise while preserving speech intelligibility, naturalness and linguistic information required by speech recognition, dialogue systems and AI agents. The solution operates entirely on-device or on edge infrastructures, without requiring cloud connectivity, making it suitable for privacy-sensitive and mission-critical deployments.
The technology has been designed for very low computational requirements and can be integrated into embedded processors, industrial devices, edge servers or telecom infrastructures.
We are looking for consortium partners developing Edge AI, industrial AI assistants, distributed AI services or AI-enabled communication systems.
CONTRIBUTION TO RELEVANT HORIZON EUROPE TOPICS
1. HORIZON-CL4-2027-DIGITAL-EMERGING-52
Human-AI voice interaction
Our technology significantly improves speech acquisition quality for conversational AI operating in noisy industrial environments.
Expected contribution:
-
improved ASR performance
-
more natural human-machine dialogue
-
robust multimodal interaction
-
improved worker acceptance
-
lower computational requirements for embedded AI
2. HORIZON-CL4-2027-DATA-03
Lightweight AI across device-edge-cloud continuum
Speech enhancement can be executed locally on embedded devices before AI inference, reducing transmitted data volume while improving inference quality.
Contribution:
-
Edge AI preprocessing
-
distributed AI pipelines
-
efficient embedded inference
-
privacy-preserving speech processing
3. HORIZON-CL4-2027-DATA-08
Distributed AI services on Telco Edge Cloud
The technology can be deployed as a lightweight speech enhancement service running on distributed edge infrastructures before cloud AI processing.
Contribution:
-
distributed speech services
-
telecom edge optimisation
-
AI-ready speech streams
-
scalable voice processing
-----------------------------------------------
KEYWORDS:
Speech Enhancement
Noise Reduction
Edge AI
Embedded AI
Industrial AI
Conversational AI
Speech Recognition
Human-AI Interaction
Telco Edge Cloud
Distributed AI
Privacy-by-Design
Low-power AI
Real-time DSP
Embedded Signal Processing
Looking for
- Product development
- Research work
- Prototype development / MVP (Product Minimum Viable)
- Licensing
Applies to
- Manufacturing
- 3C Networks
- AI-GenAI /Data/Robotics
- Other
Attached files
Organisation
Similar opportunities
Product
Real-Time Privacy-Preserving Voice Anonymisation for Trustworthy AI
- Licensing
- Research work
- Manufacturing
- Virtual Worlds
- Energy Efficiency
- Product development
- AI-GenAI /Data/Robotics
- Quantum and High Performance Computing
- Prototype development / MVP (Product Minimum Viable)
Jean-Luc CREBOUW
CEO / CTO at CandyVoice
Strasbourg, France
Product
Real-Time Audio Deepfake Detection for Trustworthy Voice AI
- Other
- 3C Networks
- Research work
- Manufacturing
- Demonstration
- Product development
- AI-GenAI /Data/Robotics
- Testing the product/application
- Prototype development / MVP (Product Minimum Viable)
Jean-Luc CREBOUW
CEO / CTO at CandyVoice
Strasbourg, France
Product
Real-Time Packet Loss Recovery for Edge Communications
- Other
- Licensing
- 3C Networks
- Manufacturing
- Product development
- AI-GenAI /Data/Robotics
- Prototype development / MVP (Product Minimum Viable)
Jean-Luc CREBOUW
CEO / CTO at CandyVoice
Strasbourg, France