Intelligent Voice Discovery by Intelligent Voice

Used by regulatory authorities, law enforcement agencies, forensic eDiscovery specialists, law firms and in-house compliance monitoring and eDiscovery teams, Intelligent Voice is the only discovery and compliance monitoring solution for unstructured audio and video content to support the entire process from ingestion and transcription through to redactions and productions of both media and text, without the reviewer or the data ever leaving the Relativity platform.

A range of published case studies and CEO endorsements speak to Intelligent Voice’s end-to-end audio and video discovery capabilities in Relativity, and its interoperability with Conceptual Analytics, Active Learning, and Machine Translation.

In addition to the player features detailed in the separate listing for the free Intelligent Audio / Video Player, transcribing media with Intelligent Voice provides a highly accurate automatically transcribed index that is speaker separated, searchable and synchronized within the media review player.

A navigable executive summary of interesting things said is provided as a sidebar in the player, with any responsive Relativity search or alert terms separately listed. These Intelligent Voice Tags, generated from a proprietary patented algorithm detailed in New Scientist magazine, serve a number of purposes:

1. An Executive Summary of the recording for assessing responsiveness, both in the Player itself and in the Workspace Document List.
2. Provide Jump To navigation for assessing key portions of the record. Who first introduced a particular topic into the conversation? Where did discussions on it wind up? Who had the final word?
3. Allow users to identify themes & concepts in the audio review, providing insight without having to review entire recordings.
4. Surface Blind Spots & Unknown Unknowns, whether that be things described more explicitly over the phone than in writing, or matters not raised at all in written communications, providing a more robust and defensible set of keywords specifically for voice.
5. Reduce variation in summarization & interpretation between different users for the same recording.
6. May enable some initially responsive items to be ruled out without audio review.

The Intelligent Voice Player supports over two dozen American, Asian and European languages, including multiple dialects and accents, as well as the ability to detect and transcribe accent and language changes mid-conversation.

Words are suggested for what might otherwise be unintelligible portions of speech, not just from the karaoke play-along through the displayed transcribed index, but also the optional sub & sur-titling of additional suggestions, along with the ability to reduce the playback rate below real-time.

The Intelligent Voice Player itself can be downloaded with all of the above functionality available in a standalone HTML file. A separate Web Video Text Track VTT file can also be downloaded.

Intelligent Voice’s AI-based cognitive computing provides a client-verified 70% reduction in costs compared with human listening and manual transcription methods for discovery, regulatory and compliance investigations, data subject access requests, and cyber breach incident response.

Available Edge either on-prem or on your private cloud of choice, or partner hosted or Secure SaaS, using Relativity Server or RelativityOne.

Intelligent Voice Discovery Workflows

• Process, search, discover, review, categorize and redact, quickly, easily and cost-effectively, without you or the data ever leaving Relativity.
• For any given matter, the lexicon can be adapted to bias towards the Search Keywords for Voice, ahead of other similar sounding words and phrases, for optimal search retrieval. A simple, sub-one hour process, all within Relativity.
• Processed audio and video content fits the same Relativity workflows as all other data for review, analysis, redactions and productions, including Relativity Conceptual Analytics and Active Learning. Start with a Relativity Search Set, conclude with one or more Relativity Production Sets.
• The ability to search identified alternatives for borderline unintelligible audio, made available search term by search term, returns highly accurate results independent of audio quality.
• Multilingual conversations can be processed in a single run and reviewed in a single item combining all spoken languages.
• Interoperability with machine language translation providers on the Relativity App Hub enables automatic translation for either search or pre-production purposes.
• Enables multiple concurrent redacted versions of the same audio file within Relativity, whether for separate review teams or multiple requesting parties, any one of which can be produced at any point.
• Also supports Data Subject Access Requests under GDPR, FOIA etc, including where a case custodian submits a DSAR mandating a response ahead of the case disclosure schedule, and with separate redactions required.
• A separate voice biometric identification capability is available prior to and outside of Relativity. It is available both for identification prior to collection and for overlaying biometrically determined speakers details as workspace metadata post ingestion, in jurisdictions where it’s permissible to both create physiological biometric data and use it to uniquely identify an individual.

Intelligent Voice Discovery Functionality

• Intelligent Voice’s ultra-high-speed GPU powered speech recognition algorithms provide results in a fraction of the time of our competitors and with no loss of accuracy.
• Intelligent Voice requires no additional installation of software on a user’s local machine, or special integration methods for so-called larger files.
• Indexing of identified alternative words overcomes the over-retrieval of traditional phonetic-based approaches, whilst also avoiding the under-retrieval of “OCR for Voice” transcriptions that record one word and one word only, regardless of any ambiguity around what may or may not have been said.
• Transcription is indexed for Keyword and dtSearch, enabling successful Conceptual Analytics and Active Learning, getting you to critical evidence quicker and giving you more time to build your case.
• Having applied redactions to both audio and text either by a simple click and drag or direct timestamp entry, users can then preview audio files with redactions applied, to ensure the file has been properly edited, prior to production.
• All features available, and in the same manner, for both Server and RelativityOne.

Intelligent Voice Discovery Pricing

• Simple, predictable cost models, with pay as you go pricing options, and no need for native file applications.
• To the cent pre-processing billing assessments and post-processing confirmations, supporting both daily and intra-day billable cost reporting.
• Documented intelligent culling processes reduce the volume of media to be hosted in the first place.
• The features available from the Intelligent Voice Media Player for untranscribed or pre-transcription media can then reduce the need for billable speech to text processing of hosted media.
• The efficiency of GPU versus CPU-based processing, and reduced hosting for processing from GPU acceleration, minimizes resource requirements for processing.
• Free trial available of all capabilities for any RelativityOne customers in the US.
• Free use for approved Relativity Justice for Change projects, leveraging the unique capabilities of Intelligent Voice and Relativity to positively impact racial and social justice in our communities.
• Free installation and use of the Intelligent Voice Media Player for untranscribed media of any amount in both RelativityOne and Server, available for all territories.

From Intelligent Voice Limited

• ISO 9001 & 27001 accredited
• Crown Commercial Service Supplier
• NVIDIA Inception Partner for Voice Analytics
• Winners of the Security Innovation Award at the UK Government’s Global Security Event