Transcribing and Categorizing 100,000 Minutes of Audio recordings

Mindy Support’s team listened to, checked, and modified audio transliteration from sales meetings. We then classified the written text into 100+ different categories and suggested improvements to existing text, which can help in future sales calls. The meetings that were transcribed were conducted in English, French, and German. 

The client was looking to optimize their sales processes by reviewing sales meetings conducted by their sales representatives over video conference. They needed all of these audio files to be transcribed into text to be later analyzed by ML algorithms, which would provide some recommendations on how to improve these sales calls. However, the sheer volume of these recordings was massive, consisting of more than 100,000 minutes of audio text. The customer was looking for a trusted data annotation partner who could transcribe all of these audio files and categorize them into more than 100 different categories. In addition to this, there was a strict deadline of six months to complete all the text transcription and categorization work. 

Mindy Support has actualized projects for clients in the past, and we have proven to the client that we understand the critical importance of accurate and reliable data annotation for their business needs. Furthermore, our commitment to excellence, coupled with our advanced technological expertise and experienced team, set us apart as the premier choice for annotation services. 

In order to complete this project with the highest level of quality, we knew that we would need onboard experts in English, French, and German. Thanks to our proven recruiting methods, we were able to hire six experienced data annotators and 1 team lead quickly and efficiently. We then implemented a special training program, which included an explanation of new terms and features that are particularly pertinent to the client’s sales process. Once all the annotation work was complete, we had a manual quality assurance process in place to ensure compliance with client requirements and maintain high standards.

In total, we transcribed more than 100,000 minutes of audio recordings and categorized each one into its relevant category. This entire process was completed with a quality score of 99%+. The client was very pleased with the stable delivery of annotated data during 6 months of the project on a monthly basis. There are future plans to expand this project to include an even larger dataset and a larger number of languages.

• 100,000 minutes of audio recording transcribed 
• Categorized the recording into 100 categories
• Achieved a 99%+ quality score