![]()
Whitehaven Coal’s fleet personnel were manually auditing Positive Radio Communications (Pos-Comms) between vehicles. This process was expensive, time intensive, inconsistent, and error-prone: auditors applied rules differently, sometimes passing or failing calls incorrectly. Missed or incomplete Pos-Comms posed real safety risks, including reduced situational awareness, near-misses, and potential collisions. Fleet personnel needed a tool to help standardise audits, reduce human error, and get actionable insights.
SSW developed a custom AI-driven auditing platform prototype that transcribes and analyses recorded radio communications from Whitehaven Coal’s mining fleet. It uses Azure OpenAI Whisper for transcription and Azure OpenAI GPT models to score compliance and identify patterns in positive communications.
The platform provides personnel with an AI-powered workflow that uploads audio files, runs AI transcriptions, reviews and corrects those outputs, and finally exports detailed reports. Visual dashboards allow for compliance gaps to be identified more clearly, supporting personnel with greater visibility and consistency within their workflow.
The planned next phase would use corrected AI outputs to refine the models and improve accuracy over time. Future phases also plan to broaden compatibility with additional radio network formats and mining sites.

Figure: Architecture Diagram
Early in development, SSW pivoted from direct database access to a unified pipeline for audio analysis due to inconsistent formats. Iterative design allowed rapid feedback from Whitehaven Coal, accommodating evolving compliance rules without disrupting progress. Ongoing testing and iterative refinement have steadily improved transcription accuracy and reliability. While still in the prototyping stage, the system continues to improve with each development cycle.