A Bash utility for analyzing objective characteristics of audio files (.wav, .flac, .mp3, etc.) using ffmpeg.
Provides quick, repeatable, file-level measurements for common audio metrics.
Most tools either:
- focus on system or room measurement (for example, REW)
- provide raw ffmpeg output without a concise summary
- output a single quality score
This tool fills that gap by providing:
- quick, objective multi-metric summaries
- a scriptable workflow for repeated analysis
- consistent results across Linux and Windows Subsystem for Linux (WSL)
- a way to download a spreadsheet of your music file data
- a way to create custom playlists of your music files based on file statistics and metadata
- Peak level (dBFS)
- Noise floor (dBFS)
- Crest factor (dB)
- Stereo correlation (phase)
- Loudness (EBU R128: LUFS, LRA, true peak)
- Metadata: genre, artist, album, track, duration, year, sample rate, average bit rate, and bits per sample can be displayed. recording-analyzer uses the metadata already in the file, it does not perform metadata lookups.
- File formats supported: aac, ac3, aif, aiff, amr, caf, flac, m4a, mp3, ogg, opus, pcm, wav, wma
The BASH shell (version 4.3 or greater), ffmpeg (version 5 or greater), and jq (version 1.7 or greater) programs must be available on your computer. On Ubuntu linux, you can install them via:
sudo apt update
sudo apt install ffmpeg jqBash is already installed on most Linux systems.
In these examples we assume that the directory $HOME/.local/bin exists and is in your $PATH. This is usually the case on most linux systems. Adjust as needed.
Download the file from the URL below and save it as $HOME/.local/bin/reca.sh
https://raw.githubusercontent.com/mcochris/Recording-analyzer/refs/heads/main/reca.sh
cd $HOME/.local/bin
chmod +x reca.shor
cd $HOME/.local/bin
curl --remote-name https://raw.githubusercontent.com/mcochris/Recording-analyzer/main/reca.sh
chmod +x reca.shor if you're familiar with the git version control system:
cd $HOME
git clone --depth 1 https://github.com/mcochris/Recording-analyzer.git
cd $HOME/.local/bin
ln -s $HOME/Recording-analyzer/reca.sh .
chmod +x reca.shreca.sh may inform you that an updated version of the program is available. You can follow the installation steps above to download the new version. If you're a git user perform a "git pull" to update the repo.
NOTE: It's important to wrap your file/directory names in double quotes to handle file names containing spaces or special characters.
reca.sh "<audio_file>"Analyze every music file in the Music directory:
reca.sh "/home/user/Music"Analyze every music file in the Music directory, including all its subdirectories:
reca.sh --recurse "~/Music"Analyze every music file in the Music directory and get each file's metadata:
reca.sh --metadata "$HOME/Music"Analyze every music file in the Music directory and capture the JSON results in a file. You can import the JSON file into your spreadsheet program, or upload the JSON file to https://recording-analyzer.mcochris.com/ to see your data in spreadsheet format, download the spreadsheet of your data, and create custom playlists:
reca.sh --metadata --json "/home/user/Music" > mydata.jsonCommand line options can be combined:
reca.sh --metadata --json --recurse "/home/user/Music"-d, --debug Enable debug mode to show detailed processing information
-e, --extensions Specify a custom list of audio file extensions to analyze.
For example, if you supply '-e "m4a flac"' on the command
line, only files with the .m4a and .flac file will be
searched for and analyzed. The default is to search for
and process all the supported extension noted in the
"Features" section above.
-h, --help Show a help message and exit
-j, --json Output results in JSON format (default: human-readable text)
-l, --limit N Limit processing to the first N audio files found
(default: no limit)
-m, --metadata Include metadata fields in output
-q, --quiet Suppress progress spinner and other non-essential output
-r, --recurse Recursively search directories for audio files
-v, --version Show program version and exit
chris@studio:~/audio$ reca.sh "Computer World.flac"
Audio Analysis: "Computer World.flac"
=====================================
Left Channel:
Peak Level: -0.50 dBFS
Noise Floor: -inf dBFS
Crest Factor: 12.30 dB
Right Channel:
Peak Level: -1.05 dBFS
Noise Floor: -inf dBFS
Crest Factor: 10.68 dB
Stereo Correlation:
Average Phase: 0.4893
Loudness (EBU R128):
Integrated Loudness: -19.37 LUFS
True Peak: -0.50 dBTP
Loudness Range: 4.00 LU
chris@studio:~/audio$ reca.sh --metadata "The Things We Do for Love.flac"
Audio Analysis: "The Things We Do for Love.flac"
================================================
Metadata:
Genre: Rock
Artist: 10cc
Album: The Very Best of 10cc
Track: 11
Duration: 213 seconds
Year: 1997
Sample Rate: 44100 Hz
Avg. Bit Rate: 887750 bps
Bits Per Sample: 16
Left Channel:
Peak Level: -1.76 dBFS
Noise Floor: -inf dBFS
Crest Factor: 6.29 dB
Right Channel:
Peak Level: -0.55 dBFS
Noise Floor: -inf dBFS
Crest Factor: 6.67 dB
Stereo Correlation:
Average Phase: 0.31
Loudness (EBU R128):
Integrated Loudness: -13.92 LUFS
True Peak: -0.53 dBTP
Loudness Range: 4.10 LU
Peak Level (dBFS) Audio peak level is the highest instantaneous amplitude or loudest point in an audio signal, measured in decibels relative to full scale (dBFS). Does not include intersample peaks. It represents the maximum transient peak, not the average loudness. Ensuring peaks remain below zero is crucial to prevent digital clipping (distortion).
Noise Floor (dBFS)
The noise floor in audio is the sum of all unwanted ambient sounds (HVAC,
traffic) and electronic hiss (preamps, interference) present in a recording
space or signal chain when no intended sound is being made. It represents
the baseline "silence" of a system; a lower noise floor allows for greater
dynamic range, while a high noise floor can make recordings sound
unprofessional. -inf dBFS indicates digital silence or values below numerical precision.
Crest Factor (dB) Crest factor in audio is the ratio of a signal's peak amplitude to its average (RMS) power, measuring the "peakiness" or dynamic range of a sound, typically measured in dB. A higher crest factor indicates high dynamics (12–15+ dB, e.g., drums), while a lower, smaller value suggests heavy compression or a denser, more consistent sound (6–9 dB).
Stereo correlation (phase) Stereo audio average phase, often visualized via a phase correlation meter, measures the similarity between left and right channels, averaging from -1 (fully out-of-phase) to +1 (fully in-phase). A positive average (+0.1 to +1) indicates good mono compatibility, while a negative average indicates potential phase cancellation, where sounds disappear in mono.
Integrated Loudness (LUFS) Audio integrated loudness is the average loudness of the audio file in LUFS (Loudness Units Full Scale). Unlike peak meters, it calculates loudness based on human perception to ensure consistent volume levels, often targeting -14 LUFS for streaming services like Spotify or YouTube.
True Peak (dBTP) Digital audio True Peak (measured in dBTP) is a standard measurement that predicts the highest level an audio signal will reach after conversion from digital to analog, accounting for peaks between samples (inter-sample peaks). While standard digital meters (dBFS) only measure sample points, True Peak uses oversampling to detect inter-sample peaks that can cause distortion when converting to lossy formats like MP3 or AAC.
Loudness Range (LRA) Loudness Range (LRA) measures the dynamic variation between the softest and loudest parts of an audio program, quantified in Loudness Units (LU). While average loudness (integrated loudness) is typically targeted at -14 to -16 LUFS for streaming, LRA ensures audio isn't too static or too erratic, with a preferred range of 5–10 LU for consistency.
You can verify the recording-analyzer's output against programs other than ffmpeg for comparison. See methodology and verification details here:
https://github.com/mcochris/Recording-analyzer/blob/main/verification/README.md
Comments, questions, and suggestions are welcome. You can open an issue here: https://github.com/mcochris/Recording-analyzer/issues