Auto-Transcription

This technical note discusses Lexbe's Auto-Transcription service.

Overview

The Lexbe Auto-Transcription service can transcribe standard 16 kHz and 8kHz audio streams and multiple audio encodings, including WAV, MP3, MP4 and FLAC.  This feature is supported by Amazon Web Services (AWS) Transcribe which is an automatic speech recognition service.  Included in this service is speaker identification.  For multi-speaker audio files--such as meetings, multi-participant phone calls, depos, etc--speaker identification can detect any change in speaker and assign the specific text accordingly.  Once the audio files have been uploaded into a Lexbe eDiscovery Platform (LEP) database, the Professional Services team can initiate the Lexbe Auto-Transcription service upon request. The transcribed text is displayed in the PDF tab of the corresponding audio file document record and is added to the case index.  This allows for comprehensive search and supports efficient review and coding.     

Benefits:

• Improved comprehensive search

• Easy to read transcriptions of audio files

• Faster review

• Speaker identification

• Single document record containing both the original audio/video file and transcribed PDF

• Underlying AWS Transcribe always learning and improving to keep pace with the evolution of language

Use Cases:

• Voicemails:  Many cloud-based communication phone systems send voicemails to the recipient via email.  As such, these can appear in a general email collection.  Voicemails can also be collected from cell phones.

• Call center recordings:  Call center recordings can be of importance in many cases including employment and business disputes.

• Recorded meetings:  Any recorded meetings saved to a Custodian’s collected location (cloud-based or local machine) will be included in the dataset so long as it is not filtered out based on specific collection criteria (search, file extension, etc.)

• Hearings/Depos/Trial:  When a transcription of these events is not available or is an added cost, the audio file can be requested and uploaded into an LEP for transcription.  This can be valuable for purposes of collaborative work and case evolution.   

• Miscellaneous email attachments: Sometimes miscellaneous audio files can be attached to emails.  Having transcribed text will speed review and determination of importance if any, of the attached file without having to listen to it in the first instance.

The Lexbe Auto-Transcription service is run on request by the Professional Services team and requires a qualifying Lexbe eDiscovery Platform hosting account.  For more information or to request this service, please contact our Professional Services team (professionalservices@lexbe.com) or Sales team (sales@lexbe.com).