The Client
Core Nova specializes in delivering fully customized SaaS solutions. They also offer customer service and marketing lead generation support across various business verticals such as banking, insurance, and automobile through its 500+ agents.
Disclaimer: The name ‘Core Nova’ is a placeholder, as there is an NDA signed between both parties.
Challenges
Core Nova faced significant issues with their existing audio detection model, requiring a faster and more accurate solution to meet their operational goals.
- Core Nova's Asterisk-based audio detection model was slow and inaccurate, achieving only 60% accuracy in 3 seconds.
- The client needed a model to detect humans or machines on calls in 1 second with 90%+ accuracy.
- Overlapping audio patterns caused difficulties distinguishing between human and machine voices, especially during the first 500 milliseconds in live environments.
- In clustering tests, 73% of Answering Machine (AM) audios and 27% of Human Answered (HA) audios fell into one cluster, showing significant overlap in audio characteristics.
- Mislabeling of audio characteristics further reduced model accuracy, making it harder to achieve the client’s goal of 90% accuracy.
Solution
To address these challenges, we implemented a comprehensive AI-driven solution tailored to the client’s needs.
- Conducted a four-week AI readiness audit to define the scope and analyze the platform’s current state.
- Filtered, organized, and labeled audio datasets for easier search and processing.
- Identified patterns to classify audio as Human Answered (HA) or Non-Human (AM) within 1 second.
- Built a Python-based predictive model capable of identifying audio characteristics within 500 milliseconds.
- Enhanced backend operations by integrating with the client’s tech stack, improving accuracy and efficiency.
- Developed a step-by-step solution:
- Dataset Creation
- Dataset Preprocessing
- Model Development & Training
- Model Assessment
- Regularly communicated with the client to validate and correctly label misclassified audio data.
Results
The solution delivered measurable outcomes, significantly improving operational efficiency and cost savings.
- Saved 30 minutes per agent daily by automating the process.
- Saved $110K per month in operating costs.
- Enabled telemarketing agents to connect with more potential clients by increasing time and bandwidth.
- Streamlined business operations with greater precision and lower operational costs without compromising service quality.
- Core Nova renewed its partnership with Maruti Techlabs for phase 2 to further enhance their product roadmap and service delivery.
