May 18, 2016 Articles

Deploying Text Analytics and Speech Recognition to Manage Audio Files in E-Discovery

Talk is the new doc.

By Peter Gronvall and Nathaniel Huber-Fliflet

Audio files—recorded conversations in corporate environments, mostly—are increasingly falling within the scope of inquiry in regulatory enforcement actions and large-scale litigations. Files of this nature can include call center recordings, trader desk headset conversations, and streaming recordings within multi-channel/trading environments.

The financial services sector in particular is facing serious challenges in how to segregate, process, review, and actually understand the content of audio files in these compulsory-process scenarios—and then, further, how to produce those files over, in native/redacted formats. What has become evident, in nearly every instance, is that the “brute force” review of audio files (i.e., minute-by-minute listening and coding of those files) is inefficient, cost prohibitive, and vulnerable to substantial human error. These challenges are similar to those the legal community faced with email during the information explosion, and just as with email, technology is helping to overcome those challenges.

An engineered solution that combines up-front human transcription, machine learning applied to speech recognition, and sophisticated text analytics is essential to move beyond brute force review. Solutions of this nature allow companies to achieve efficiency, scale, accuracy, and substantial cost savings.

Premium Content For:
  • Litigation Section
Join - Now