Customer Support Center: Enhanced Analytics
Challenge
A large sportswear retailer* wanted to identify which products customers were having issues with. To do this, the retailer needed to analyze mentions of brands and models in phone calls, along with the context, to determine customer satisfaction with their purchases and the quality of service.
Due to the company's security policy, cloud solutions were prohibited, and existing local speech transcription options could not recognize brand names in the required languages.
*The company's name is not disclosed in compliance with confidentiality agreements.
Solution
Product: On-Premise Speech Recognition Software
The Lingvanex team first requested a list of product names from the client for analysis. They then created a specialized dataset, combining dictated audio and corresponding text data that included the names of sportswear brands and models.
This dataset was used to train a language model capable of automatically transcribing telephone calls into text, which could then be seamlessly integrated into the client’s software for automatic analysis.
Results
The solution transcribes tens of thousands of hours of phone calls daily, accurately identifying brands and product models across more than 90 languages.
To protect user data, the language model is deployed locally on the client’s infrastructure. This allows for easy integration into the client’s analytical services via the REST API, ensuring seamless and secure processing of customer interactions.
Contact Us
Completed
Your request has been sent successfully