CandyVoice is a French DeepTech company developing advanced AI-powered voice technologies combining artificial intelligence (AI) and digital signal processing (DSP) to enable secure, trustworthy and computationally efficient voice applications.
Our software solutions are designed to operate in real time on standard CPUs, embedded systems, ARM processors, DSPs, edge devices or cloud infrastructures, without requiring GPU acceleration or permanent cloud connectivity. Their low computational footprint, low latency and hardware efficiency make them particularly well suited for next-generation Edge AI, distributed AI and Telco Edge Cloud infrastructures.
CandyVoice develops a portfolio of complementary technologies covering the entire voice processing chain:
• Voice Anonymization providing irreversible protection of biometric voice identity while preserving speech intelligibility and compatibility with automatic speech recognition.
• AI Noise Reduction significantly improving speech intelligibility in highly challenging acoustic environments.
• Lost Audio Packet Recovery reconstructing consecutive missing voice frames to maintain communication quality over degraded or high-latency networks.
• Audio Deepfake Detection detecting AI-generated or manipulated speech in real time or offline, with or without a reference voice.
• Real-time Voice Imitation enabling highly realistic multilingual voice avatars, immersive communications and dynamic voice identities with minimal training data.
Our technologies address several strategic application domains, including telecommunications, Edge AI infrastructures, mission-critical communications, cybersecurity, privacy-preserving AI, trusted digital identities, industrial digitalisation and next-generation virtual worlds.
For immersive applications, CandyVoice enables natural multilingual voice avatars, conversational AI, digital twins, XR collaboration platforms, simulation and training environments. Voice anonymization and deepfake detection contribute to trusted virtual interactions, while real-time voice imitation creates natural and expressive user experiences for human-centric AI systems.
The company places particular emphasis on European strategic priorities including trustworthy AI, digital sovereignty, cybersecurity, privacy by design, computational efficiency and sustainable AI. Our technologies are designed to support decentralised AI processing, low-latency services and resilient digital infrastructures where local execution, energy efficiency and data sovereignty are essential.
In June 2026, CandyVoice received international recognition by winning the International Critical Communications Award (ICCA 2026) in the "Best Use of Advanced Technology" category for its Mission-Critical AI Voice Stack. Selected by an international jury of 60 experts, this award recognises the innovation and operational value of our AI voice technologies.
Based in Strasbourg, France, CandyVoice is actively seeking industrial companies, telecommunications operators, research organisations, universities and system integrators to participate in Horizon Europe collaborative projects related to Edge AI, trustworthy AI, decentralised AI processing, Telco Edge Cloud infrastructures, immersive virtual worlds, cybersecurity, resilient digital infrastructures and advanced AI applications.