Audio & AI Transcription
Inspecting a hive often leaves your hands full. Hive-Pal lets you speak your observations and have them transcribed and turned into structured inspection data automatically — so you can keep your eyes on the bees.
When to Use It
- You want a hands-free way to capture observations in the field.
- You'd rather talk through an inspection than tap through a form.
- You want a voice record attached to the inspection for later reference.
Recording Audio
There are two ways to record:
- Audio Quick Inspection (
/hives/:id/inspect/audio) — a streamlined page focused on recording a voice note for a hive. - Within an inspection — the standard inspection form has an audio section where you can attach a recording.
When you save, the recording is uploaded with the inspection.
How Transcription & Analysis Work
Processing happens in two stages after upload:
- Transcription — a worker converts your recording to text. The audio card shows the status: waiting → transcribing → completed (or failed).
- AI analysis — once transcribed, you can trigger analysis. The AI reads the transcript and suggests values for inspection fields (ratings, observations, actions).
You stay in control: review the transcription (edit it if needed), then merge the AI's suggestions into the inspection form. Nothing is saved to your inspection until you accept it.
Requirements
- File storage must be enabled (local or S3) to store recordings — see Configuration → File Storage.
- The transcription/analysis worker must be running for processing to complete.
FAQ
Is my audio kept? Recordings are stored in your configured file storage (your own server with local storage, or your S3 bucket). As a self-hosted app, the data stays under your control.
Can I edit the transcription? Yes. The transcribed text is editable before you run analysis or merge it into the inspection.
Do I have to accept the AI's suggestions? No. Suggestions are proposed as mergeable values — you choose what to apply to the inspection.
What if transcription fails? The audio card shows a failed state; you can retry. Check that the worker is running and file storage is configured.