Fire scenes change as they are made safe, examined and cleared. Debris is moved, access improves and early explanations are revised. A useful record must show what the investigator observed at each stage, what witnesses reported and how competing hypotheses were tested.
An AI voice recorder for fire investigators can support authorised scene narration, witness interviews and technical debriefs. Its purpose is to preserve contemporaneous detail. It must not convert a provisional idea into a final origin or cause conclusion.
Use a spoken scene slate before every recording
Begin with:
- incident and scene reference
- date and time
- investigator
- location within the scene
- people present
- scene status and safety restrictions
- purpose of the recording
If the recorder is paused and resumed elsewhere, create a new location and time marker. This makes later review far more reliable.
Narrate from general to specific
A disciplined scene narration may move through:
- external approach and building context
- entry route and condition
- room or area sequence
- patterns and physical effects
- potential ignition sources
- fuel and ventilation factors
- electrical, mechanical or appliance evidence
- photographs, samples and exhibits
- limitations and areas not examined
Describe what is visible before interpreting it. This preserves the source observation if the hypothesis later changes.
Use three labels throughout the examination
| Label | Meaning |
|---|---|
| Observation | A condition directly seen or measured |
| Information | A fact or account supplied by another person or record |
| Hypothesis | A possible explanation requiring testing |
Example:
“Observation: localised damage to cable insulation behind appliance. Information: occupier reports appliance was running before discovery. Hypothesis: electrical involvement remains possible; preserve component and test against alternative heat-damage explanation.”
This is safer than summarising “electrical fault caused the fire”.
Build a hypothesis tree, not a single story
For each credible hypothesis, record:
- proposed area or point of origin
- possible ignition mechanism
- fuel and fire-development explanation
- evidence supporting it
- evidence inconsistent with it
- tests or specialist examination required
- alternative hypotheses
- current status: active, weakened, excluded or unresolved
An AI system may favour the explanation repeated most often. The investigator should preserve all reasonable alternatives until the evidence supports their treatment.
Connect every spoken observation to an evidence reference
Read out:
- photograph and video numbers
- sketch or plan references
- sample and exhibit identifiers
- measurement records
- document or log references
- specialist examination requests
Repeat identifiers and verify them against the evidence register. The transcription must not create exhibit numbers or correct an uncertain identifier by guesswork.
Witness interviews: preserve sensory detail and uncertainty
Where interviews are authorised, ask what the witness personally saw, heard, smelled or did. Establish location, timing and line of sight before testing details.
Record:
- the witness’s position
- sequence of events
- first indication of fire
- doors, windows and ventilation
- equipment operating
- actions taken
- people present
- uncertainty and estimates
- information learned from others
A witness’s interpretation should not replace their direct observation. Preserve “I thought it came from the cooker” separately from “I saw flames at the rear of the cooker”.
Scene alteration must be documented
Record movement or changes caused by:
- firefighting
- rescue
- making the scene safe
- weather
- structural failure
- investigative access
- removal of exhibits
Where possible, state who made the change, why and when. This helps later reviewers understand the condition of the scene at each stage.
A source-to-report workflow
- Confirm recording authority and scene safety.
- Create the spoken scene slate.
- Narrate location and observation before interpretation.
- Reference photographs, samples and exhibits.
- Maintain the hypothesis tree.
- Record limitations and scene alteration.
- Generate the transcript through the approved process.
- Correct technical terms and evidence identifiers.
- Compare the transcript with images, plans, tests and witness material.
- Prepare the formal report with clear reasoning and uncertainty.
Do not let transcription outrun safety or evidence control
The device must not distract from PPE, structural hazards, contamination controls or communication with the scene team. In unsuitable areas, use brief identifiers and complete a structured debrief in a safe location.
Source audio, transcripts and extracts should follow the investigation’s access, disclosure and retention process.
How NERALVO Halo may support authorised scene work
NERALVO Halo provides portable NOTE recording, 64GB local storage and DOWAY transcription tools. It may support approved scene narration, witness interviews and post-scene debriefs.
Test PPE, wind, reverberation and technical vocabulary before operational use. Evidence handling and file naming should be defined before the scene. Structural-site observation methods are also covered in the structural engineer guide.
Investigation-record check
- Does every recording have a scene slate?
- Are observations separated from information and hypotheses?
- Can evidence references be verified?
- Are alternative hypotheses visible?
- Are witness interpretations separated from sensory observations?
- Are scene alterations and limitations documented?
- Does the final conclusion follow tested evidence?
An AI voice recorder can improve the completeness of a fire-investigation record. The reliable conclusion still depends on disciplined scene methodology, evidence continuity and hypotheses that are tested rather than merely narrated.
Ready to capture meetings properly?
View the NERALVO Halo AI voice recorder with 64GB local storage, meeting capture, compatible phone-call recording workflows and one year of DOWAY Max included.
View NERALVO Halo