Text & Audio Summarisation
Overview
A single case can hold hundreds of thousands of messages, transcripts, and screens of text. Finding the ones that matter early — not months in — can change its outcome.
Rigr AI’s Text & Audio Summarisation capability securely condenses transcripts and on-screen text — across languages — into concise summaries suitable for rapid triage and evidential preparation.
When AI Invents the Answer
There are two ways a machine can summarise, and each has a hard limit.
Extraction lifts sentences straight from the source. Every word is the original, exactly as it was written or spoken — so you can trust it absolutely. The drawback: the source may never have stated, in any single passage, the precise thing you need to know.
Abstraction writes a fresh summary in its own words. It can answer your question directly, draw together threads from across a long document, and render the answer in a language other than the original. The drawback: those words are generated by an AI model, and models can hallucinate — so a fluent, confident summary may have no basis in the source material.
A Fast Answer You Can Check
The latest version of our Summariser gives you an abstract — a direct, readable answer in plain language, across languages where needed — and backs every claim with an extract from the original text.
When the abstract is right, you have a fast answer. When it draws the wrong conclusion — and any abstractive system sometimes will — the supporting extract sits right beside it, in the source’s own words. The reviewer always sees both, so a flawed conclusion can never pass as evidence unchecked.
This is summarisation built for evidential work, not just the inbox: the speed of abstraction, with the auditability of extraction.
Operational Use
- Rapid triage of long transcripts and on-screen text
- Multilingual translation and summarisation
- Identification of key statements and entities
- Every summary claim traceable to its supporting passage
Designed for sensitive and challenging content that general-purpose AI systems often reject.
Deployment and Control
- Fully containerised
- On-premise and air-gapped deployment
- No external data dependency
- Customer retains control of all data
Reviewing bodycam, interviews or jail calls?
Source-linked summarisation is what makes AI output usable in evidence review: every generated claim stays attached to the passage it came from. See how it applies to audiovisual discovery in criminal defense video evidence.
Frequently asked questions
How does Rigr AI avoid AI hallucination in summaries?
It pairs abstraction with extraction: every claim in the readable abstract is backed by a verbatim extract from the source, so a flawed conclusion can't pass as evidence unchecked.
Can it summarise across languages?
Yes. It summarises and translates transcripts and on-screen text, rendering the answer in a language other than the original where needed.
Is it suitable for sensitive content?
Yes. It is designed for sensitive and challenging content that general-purpose AI systems often reject, and runs fully containerised on-premise or air-gapped.
Can it handle a large case?
Yes. Within VST Teams it summarises each transcript and document in a case individually, so investigators can triage item by item rather than reading everything in full — every summary backed by its source extracts.