Back to Posts

Automated Voice Processing with Artificial Intelligence at Scale: RepsMate Success Story

In cutting-edge technology, National Competence Centers (NCCs) are at the forefront of groundbreaking collaborations that pave the way for innovations in artificial intelligence (AI) and data analytics. This blog post delves into the success story of RepsMate powered by NCC Romania, hosted by the National Institute for Research & Development in Informatics - ICI Bucharest, and NCC Netherlands, hosted by SURF.

RepsMate, a key player in this collaborative effort, is pioneering a SaaS solution that integrates AI and Data Analytics technologies seamlessly. The primary goal is comprehending customers' needs and behaviours and empowering company agents with automated and tailored guidance. This virtual coach for agents utilizes automatic speech recognition, spoken emotion recognition, behavioural segmentation, and real-time contextual guidance, all powered by advanced machine learning technologies.

The Technical and Scientific Challenge

The technical solution presented by RepsMate poses a unique challenge in its optimal implementation and the cohesion of its various elements, ensuring seamless operation on large-scale computing machines. Using automatic speech and emotion recognition, coupled with behavioural segmentation, demands rigorous testing and optimization for large-scale efficiency.

Collaborative Solutions

NCC Romania took the lead in analyzing RepsMate's audio processing workflow and explored possibilities for parallelization on Graphics Processing Units (GPUs). This involved an initial screening and evaluation to identify the application's potential for supercomputing power. Subsequently, RepsMate developers gained access to the Dutch national supercomputer Snellius, leveraging Nvidia GPUs for medium-scale testing and receiving crucial performance evaluation assistance from NCC Netherlands.

Scalability and Business Impact

The successful analysis showcased RepsMate's scalability and efficient utilization of large-scale systems with Nvidia GPUs. Further examinations are underway to assess scalability on AMD GPUs. This collaborative effort is a testament to RepsMate's justification for supercomputing power in handling the parallelization requirements of deep learning-based AI solutions. The business impact is significant, with RepsMate proving to be a well-implemented code capable of optimizing HPC systems.

Tangible Benefits

The collaboration between NCCs and RepsMate has yielded tangible benefits, including:

  • Concrete Application Insight: A detailed understanding of how RepsMate operates on large-scale computing systems.
  • Optimal Processing Verification: Verification of the optimal processing methodology for the application.
  • Efficient Resource Utilization: Demonstrated efficient use of computing resources in medium-scale test runs.
  • Guidance for Future Use: Invaluable guidance for RepsMate's utilization of Europe's largest EuroHPC pre-exascale systems.

In conclusion, this collaboration exemplifies the successful integration of AI technologies in voice processing, showcasing the potential for efficient and scalable solutions that benefit businesses and consumers alike. RepsMate's journey is a testament to the transformative power of collaborative efforts in advancing technology at scale.

Context:

NCC Romania operates under the umbrella of the National Institute for Research & Development in Informatics (ICI Bucharest), a pivotal institute in Romania dedicated to research, development, and innovation in information and communication technology (ICT). Meanwhile, NCC Netherlands finds its home at SURF, an ICT cooperative association supporting Dutch educational and research institutions in pursuing optimal digital services.

Subscribe to our newsletter

Subscribe to our newsletter. Stay updated on training events and latest news.

Thanks for joining our newsletter.
Oops! Something went wrong.