Skip to content
MediaWR
Wraith / 03

VOX

Voice, understood.
The VOX transcript view. In the Feel The Love project, the transcript 250829_DAY1_CLARE_AND_DOM_DATE_1600 shows the conversation as speaker-labelled segments between Dom and Clare, each with a timecode range and a confidence score. A filter panel counts 76 segments for Clare and 79 for Dom, and a properties panel summarises the recording and lists its key moments.

VOX transcribes recorded speech, identifies who is speaking, and turns it into a project you can search and question.

Powered by MediaWR's FusionASR technology, VOX is tuned for the hard cases: unscripted, multi-speaker production audio. Every speaker is given a globally unique voice fingerprint, recognised wherever that voice appears, from a single shoot day to an entire archive. Transcripts, notes, stories and metadata stay bound to the media they came from, so the whole production can be questioned in plain English, and every answer cites its source, down to the timecode.

Why VOX

  • FusionASR

    VOX is transcribed by FusionASR, MediaWR's own speech recognition engine. It recognises the speech, works out who is speaking and timestamps every word, and returns one speaker-aware transcript. More about FusionASR.

  • Persistent voice fingerprints

    One fingerprint database can span everything you run, or be scoped to a single project. Either way, name a speaker once and that identity follows them through interviews, shoot days and archives, building a history as the library grows.

  • Ask your project

    Ask in plain English. VOX searches transcripts, notes, stories and metadata together, and an answer can draw on moments from several recordings at once.

  • Story mode

    Play a recording and pin notes to the timecode as you go. Notes gather into stories, and those stories become searchable material like everything else: ask about one and VOX connects your notes with the corresponding transcript and media.

  • Runs on your own hardware

    Deploy VOX on your own compute and keep media, transcripts and speaker data inside your environment. No SaaS, no per-minute charges, and nothing leaving your network.

  • Part of the Wraith suite

    VOX is tightly integrated with the rest of the Wraith suite. REALTIME can ingest live sources straight into VOX, and FORGE workflows can ingest recorded media automatically.

The People view in VOX. Eight people — Alan, Anna, Clare, Dom, Frank, Harry, Jonathan and MIV Producer — are listed with the number of voice fingerprints attached to each. The selected person, Frank, has two fingerprints with match scores of 99% and 100%, and an Attach fingerprint action.
Name a voice once — the fingerprint finds it everywhere else.

Ask your production

Ask VOX about a contributor, storyline, theme or event and it searches across the project, connects the relevant moments and builds an answer. Every finding links back to the recording, the transcript and the timecode it came from.

Use it to investigate how a story developed, compare what contributors said across different recordings, find earlier context for a later event, or explore an idea without knowing the exact words that were used.

VOX's Chat view answering a question in plain English. Asked to find contestants who have overcome adversity, for soundbites to tie together a story about Dom and Clare, VOX suggests quoted lines from their recordings — each cited with the media file and timecode it came from, such as 250829_DOM_MIV_210830 at 00:09:48:17.
Asked in plain English; answered in soundbites, each linked to its recording and timecode.

Use cases

  • Story and theme discovery

    Explore storylines, themes and ideas across an entire production. Ask editorial questions in natural language and find relevant material even when the same idea was expressed in different words.

  • Searchable rushes and recordings

    Timecoded, speaker-aware transcripts of everything you shot, ready for review, logging and the edit. Search them by person, subject, phrase or meaning, with every result linked back to the source media and timecode.

  • Contributor histories

    Follow what a contributor has said across interviews, shoot days and source files. Find earlier context, changing positions, contradictions and related conversations.

  • Editorial continuity

    Story mode keeps producer notes, logger observations, transcripts and source material connected, so knowledge developed during production carries forward into post-production and the edit.

  • On-set research and interview preparation

    Review what contributors have said across previous recordings while filming is still underway. Use their wider recorded history to prepare follow-up questions and investigate developing stories.

  • Archive intelligence

    Add transcription, speaker identities and conversational search to existing collections, turning recorded content into something you can search and question without trawling through it by hand.

Bring us your material

VOX is easiest to judge on material you already know. Send us a representative recording and a few questions your team asks, and we'll show you what VOX finds in it.

Book a demo