RTL-SDR FM RDS Recorder

Overview

Use this skill for reliable broadcast FM recording from an RTL-SDR. The preferred workflow is IQ capture with rtl_sdr, RDS station-name extraction, offline WBFM demodulation, and MP3 export. This skill currently targets Linux environments and should be treated as Linux-only.

Workflow

1. Run --check to confirm the RTL-SDR is available and reachable before promising the skill is ready to use.
2. Confirm redsea is installed and reachable, either from PATH or from the expected local fallback path, before promising full RDS-ready behavior.
3. If redsea is missing, rebuild or reinstall it before running --decode-rds or any staged recording flow that depends on station naming.
4. When the user wants to discover stations first, run broadcast FM band scan mode and return a list of candidates with frequency and relative strength. Do not promise station names from scan mode alone.
5. Use scripts/fm_iq_pipeline.py for recording.
6. If needed, read references/pipeline-notes.md for setup-specific caveats.
7. Prefer a staged workflow: scan first, then decode RDS for one chosen frequency, then record that same chosen station with --record --freq, using the same --out-dir for both steps.
8. Return the produced MP3 file(s) to the user and mention station name / frequency.
9. For Telegram file delivery, send media-only replies when attaching generated files. Do not mix explanatory text with the MEDIA:... token in the same message; send the file attachment as its own message. If one reply contains multiple MEDIA:... tokens, put each token on its own line.

Dependencies