Stop paying humans to listen to corrupted audio files.
Fix them automatically.
Sonic Gate is a CLI-first audio/video quality gate that uses deterministic audio analysis to catch corrupted, invalid, or low-quality media files before they reach human reviewers or downstream pipelines.
Unlike tools that rely on slow, unreliable AI for everything, Sonic Gate uses fast, accurate signal processing (LUFS, silence detection, format validation) and only optionally adds AI probing when you explicitly want it.
Measure integrated loudness using FFmpeg ebur128. Ensure your audio meets broadcast standards.
Find dead air, trailing silence, and gaps using pydub. Configurable thresholds.
Detect corrupted files, wrong codecs, and unsupported formats before they crash your pipeline.
Auto-extract audio from MP4, MOV, AVI, MKV, WebM for analysis without manual conversion.
Auto-trim silence, normalize LUFS, remove DC offset. Non-destructive output to ./fixed.
Whisper-based speech detection and language validation. Off by default. Install with pip install "sonic-gate[ai]".
# Install
pip install sonic-gate
# Analyze a file
sonic-gate interview.wav
# Analyze a directory
sonic-gate ./recordings/
# JSON output for CI
sonic-gate --format json ./files/ > report.json
# Fix failed files
sonic-gate --fix ./recordings/Fast: ~4ms per file for traditional analysis. No GPU needed.
Deterministic: Same input, same output. No random AI hallucinations.
Language-agnostic: Works with any audio — English, Kinyarwanda, Tibetan, silence, music. No training data bias.
CI-friendly: Exit code 1 on failure. JSON output. Configurable rules.