How to use the wav level analyzer
- Choose a WAV file or drop it on the box. It can be up to 25 MB. Uncompressed 8, 16, 24 and 32-bit PCM and 32 or 64-bit floating point files with 1 to 8 channels are read.
- Wait a moment while the file is measured on your device. Large files take a second or two.
- Read the duration, sample peak, RMS level, headroom, clipped samples, DC offset and peak to RMS ratio, then the table for each channel and the table of levels through the file.
- Use Copy report or Download .csv to keep the numbers.
What the numbers mean
All levels are measured against full scale, which is 0 dBFS, the loudest value the format can store. The sample peak is the largest sample, and the RMS level is the square root of the average squared sample, a measure of how loud the file is on average. A pure sine wave has an RMS level 3.01 dB below its peak.
Headroom is how far the peak sits below full scale. A file with no headroom may have been clipped, and the page counts samples stored at the largest or smallest value the format can hold, and the number of separate runs of such samples. A few isolated clipped samples may not be audible, while long runs usually are. A file made by clipping a loud recording often shows many runs.
An example with known answers
For a test file with 8,000 frames at 8,000 samples per second, the left channel holds a 440 Hz sine at half of full scale. The page reports a left peak of -6.02 dBFS and an RMS level of -9.03 dBFS. The right channel is a steady value of 1000 with 100 samples at the top value and 3 at the bottom value, so its peak is 0 dBFS, with 103 clipped samples in 2 runs. Across both channels the RMS level is -11.59 dBFS and the DC offset is 2.11% of full scale. The same numbers came out of a separate calculation with numpy.
A DC offset is a constant shift of the whole signal from zero. A small one is harmless, but a large one wastes headroom and can cause clicks when the file is edited.
Limits
This is a sample peak, which can miss peaks that occur between samples when the sound is converted to analog or compressed, and RMS is not a loudness standard such as LUFS. The file is read once and processed on your device, and nothing is uploaded. Compressed files (MP3, AAC, FLAC) are not read; convert them to WAV first. To cut a part out of a WAV file, use the WAV Trim and Fade Tool; to change its speed, the Audio Speed Changer; and to turn levels into ratios, the Decibel Calculator.
More file tools
For audio cutting and PDF page sizes, try the WAV Trim and Fade Tool and PDF Page Size Inspector.
Assumptions and limits
- The file is an uncompressed WAV file with PCM integer or floating point samples.
- Full scale is 1.0 for floating point and the largest code for integer samples.
- A clipped sample is one stored at the largest or smallest code of the format, or at or beyond plus or minus 1.0 for floating point.
- The file stays on your device. Nothing is uploaded or stored.
- The sections divide the file into up to ten equal parts.
Frequently asked questions
What does dBFS mean?
It means decibels relative to full scale. 0 dBFS is the loudest value a digital file can hold, and every other level is negative, such as -6 dBFS for half of full scale.
Why do I see "silence"?
If every sample is zero, the level in decibels is not defined, so the page says silence instead of a number.
Does it measure loudness like streaming services do?
No. Those services use a loudness measure called LUFS that weights frequencies and averages over time. This page reports peak and RMS only.
Why does it refuse my file?
The most common reasons are that it is not a WAV file, that it is compressed, that it has more than eight channels or that it is over 25 MB. The message says which.