rms-scan

rms-scan is a small offline CLI that analyzes a media file with FFmpeg astats and reports:

• Overall/average RMS: lavfi.astats.Overall.RMS_level (dBFS)
• Highest RMS window: lavfi.astats.Overall.RMS_peak (dBFS)
• Pass/fail against a configurable RMS window (default: [-23, -18] dBFS)

It also prints per-channel RMS details when available.

Why / use cases

• Broadcast/podcast loudness spot-checks (quick RMS window compliance checks)
• Batch validation of large media libraries
• CI gating (fail builds when audio is consistently too quiet/loud)

Requirements

• Python 3.10+
• FFmpeg + FFprobe in PATH (or pass explicit paths via --ffmpeg / --ffprobe)

Install FFmpeg:

• macOS (Homebrew): brew install ffmpeg
• Debian/Ubuntu: sudo apt install ffmpeg

Install

From project root:

pip install .

After install, the console script is available as:

rms-scan --help

You can also run directly:

python3 rms_scan.py --help

Usage

rms-scan <path-to-audio-or-video-file>

Options:

• --min minimum RMS threshold in dBFS (default: -23)
• --max maximum RMS threshold in dBFS (default: -18)
• --json machine-readable output
• --verbose print raw astats lines being parsed
• --ffmpeg <path> override ffmpeg binary
• --ffprobe <path> override ffprobe binary

Example (human output)

$ rms-scan program.wav
File: program.wav
Duration: 3728.41 s
Audio: 48000 Hz, 2 ch, stereo
Overall RMS_level: -21.34 dBFS
Overall RMS_peak: -18.62 dBFS
Spec window: [-23.0, -18.0] dBFS
Result: PASS
Details:
  Channel 1 RMS_level: -21.20 dBFS
  Channel 1 RMS_peak: -18.55 dBFS
  Channel 2 RMS_level: -21.48 dBFS
  Channel 2 RMS_peak: -18.70 dBFS

Example (FAIL + suggested gain)

$ rms-scan quiet_mix.mp3
File: quiet_mix.mp3
Duration: 912.03 s
Audio: 44100 Hz, 2 ch, stereo
Overall RMS_level: -24.10 dBFS
Overall RMS_peak: -20.80 dBFS
Spec window: [-23.0, -18.0] dBFS
Result: FAIL
Suggested gain change: +3.6 dB
Details:
  Channel 1 RMS_level: -24.05 dBFS
  Channel 1 RMS_peak: -20.77 dBFS
  Channel 2 RMS_level: -24.15 dBFS
  Channel 2 RMS_peak: -20.84 dBFS

Suggested gain change targets midpoint -20.5 dBFS using:

target - measured_rms_level

Example (--json)

{
  "channel_layout": "stereo",
  "channels": 2,
  "details": {
    "per_channel": {
      "1": {
        "RMS_level_dbfs": -21.2,
        "RMS_peak_dbfs": -18.55
      },
      "2": {
        "RMS_level_dbfs": -21.48,
        "RMS_peak_dbfs": -18.7
      }
    }
  },
  "duration_seconds": 3728.41,
  "file": "program.wav",
  "overall": {
    "RMS_level": {
      "last_dbfs": -21.34,
      "max_observed_dbfs": -21.34,
      "selected_dbfs": -21.34
    },
    "RMS_peak": {
      "last_dbfs": -18.62,
      "max_observed_dbfs": -18.62,
      "selected_dbfs": -18.62
    }
  },
  "pass": true,
  "range": {
    "max_dbfs": -18.0,
    "min_dbfs": -23.0
  },
  "sample_rate_hz": 48000,
  "suggested_gain_change_db": -0.84,
  "target_midpoint_dbfs": -20.5
}

Exit Codes

• 0: PASS (Overall RMS_level in range)
• 1: FAIL (outside range)
• 2: ffmpeg/ffprobe missing
• 3: astats parse failure
• 4: invalid input file

FFmpeg Analysis Method

The tool uses FFprobe first for metadata and then analyzes audio without rendering output:

• disables non-audio streams: -vn -sn -dn
• null muxer output: -f null -
• astats filter with metadata keys:
  • preferred: astats=metadata=1:reset=0:measure_overall=RMS_level+RMS_peak:measure_perchannel=RMS_level+RMS_peak,ametadata=print
  • fallback for older builds: astats=metadata=1:reset=0,ametadata=print

This processes as fast as decode/filter speed allows (typically faster-than-real-time on modern machines).

Troubleshooting

Error: Unable to find ffmpeg/ffprobe

• Install FFmpeg package (brew install ffmpeg or sudo apt install ffmpeg)
• Or pass binary paths with --ffmpeg and --ffprobe

Parse failure (exit code 3)

If lavfi.astats.Overall.RMS_level / RMS_peak are not found:

• Re-run with --verbose to inspect parsed lines
• Check filter availability:
  • ffmpeg -filters | grep astats
  • ffmpeg -filters | grep ametadata
• Try a newer FFmpeg build (some builds vary in filter/metadata behavior)

Input fails validation (exit code 4)

• Verify file exists and contains an audio stream
• Confirm ffprobe can read it:
  • ffprobe -v error -show_streams -show_format <file>

Development

python -m venv .venv
source .venv/bin/activate
python -m pip install -U pip
python -m pip install -e ".[dev]"
python -m unittest discover -s tests -p "test_*.py"
python -m build
python -m twine check dist/*

Release

1. Update pyproject.toml version and CHANGELOG.md.
2. Build and check:
   • python -m build
   • python -m twine check dist/*
3. Create and push a tag like v0.1.0. The GitHub Actions release.yml workflow publishes to PyPI via Trusted Publishing (OIDC).