Free tools for your next audio edit
Adjust a recording toward your chosen integrated loudness with one gain across every channel. A −1 dBFS sample-peak ceiling can constrain that gain. Compare the input and actual exported WAV measurements before downloading; your audio stays in this browser.
Fixed sample-peak ceiling: −1 dBFS. One gain changes the whole recording. If peaks prevent enough gain, the exported loudness stays below the requested target. No limiter or true-peak measurement is used.
No account, credits, or audio upload. This is not compression, limiting, noise removal, or a complete broadcast/streaming compliance check. The output must fit within 50 MiB. Save it before leaving.
Measure, adjust, verify
Keep the best original. Use the complete recording you intend to measure, within the tool's limits.
Choose mono or stereo audio from 0.4 seconds to five minutes, up to 50 MiB. The browser decodes at 48 kHz. A selected file stays in this browser; processing does not start an AI job or require an account.
Enter −30 to −5 LUFS. The default −16 LUFS is a starting value, not a delivery preset. One linked gain adjusts the complete recording while a fixed −1 dBFS sample-peak ceiling can restrict the gain. The tool does not compress or limit peaks to force the target.
Prepare the result, then compare input LUFS, requested target, applied gain and exported LUFS/sample peak. Listen to the original and actual WAV. If peak headroom prevented reaching the target, the report says so. Download only the version that suits the edit.
A different target from sample peak
The meter uses K-weighted energy, 400 ms blocks with 100 ms hops, and absolute and relative gating for supported mono/stereo audio. It measures the complete decoded recording, not a waveform thumbnail or only the loudest instant. The EBU loudness-metering reference explains the broader metering framework; this utility is not a complete EBU-mode or delivery-compliance meter.
Requested gain in dB = target LUFS − measured input LUFSFor example, an input at −22 LUFS needs a requested +6 dB to approach −16 LUFS. If its sample peak is −10 dBFS, that boost can stay below the −1 dBFS sample ceiling. If its peak is already −2 dBFS, roughly +1 dB is available instead. The tool uses the smaller gain and reports the measured output rather than declaring the requested target achieved.
The ceiling is rounded conservatively for PCM16 sample quantization. Every channel receives the same gain, preserving relative balance before integer rounding. Quiet passages remain quiet relative to louder passages; this is not word-by-word leveling or a dynamics compressor.
Choose by the result
Listen in the actual project, and keep the original. Equal integrated loudness readings do not guarantee that every moment, timbre, or listener's experience is identical. A short excerpt also does not establish a longer programme's loudness.
A short loud impact can leave too little headroom for a larger constant boost. The result may stay quieter than requested. Choose a lower target or deliberately address dynamics in an appropriate editor; this tool does not hide clipping behind a successful target label.
Silence or very quiet audio can have no complete blocks above the absolute gate. A meaningful LUFS target cannot be calculated from that result. Check the original and its recorded channels rather than repeatedly boosting a silent file.
The export is the evidence
The report measures the integer samples written into the actual WAV, not only the theoretical gain. The fixed-gain calculation can differ slightly from the resulting integrated loudness because absolute gating and integer quantization can change which material contributes. The original file remains unchanged.
Output is 48 kHz, 16-bit PCM WAV with the full decoded frame count and original mono/stereo channel count. There is no trim, fade, noise removal, channel remix, or metadata copy. Both input and output must fit within 50 MiB; a compressed input can produce a larger WAV. Use the file size calculator or trimmer to prepare a suitable working copy.
The sample ceiling does not measure or guarantee true peak between samples. Later mixing, effects, resampling or MP3 encoding can change peaks and loudness. Measure the actual final delivery with tools that cover its required specification.
Download each original tone example, choose it here separately, and compare the measured outputs at the same requested target. A file is not transferred automatically.
Pick the quantity you need
The sample-peak normalizer sets a peak target and does not target LUFS. The LUFS meter measures without changing audio. This loudness normalizer requests a LUFS target and remeasures its adjusted WAV. Choose the operation your project needs rather than assuming the three names mean the same thing.
Raising the recording also raises noise already present. For a competing source, use the cleanup-method comparison before a final level adjustment. A louder version is not evidence that noise was separated.
A few useful answers
Choose a supported mono or stereo MP3/WAV, enter a target from −30 to −5 LUFS, and prepare the WAV. The tool measures the input, applies one gain constrained by a −1 dBFS sample-peak ceiling, then measures the actual exported integer PCM. Use the reported output, not the requested target, as the result.
Yes. Processing, previewing, and WAV download are free, with no account, credits, or audio upload. They run in this browser. The separate AI isolation service has its own account and download-credit requirements. Download your result before leaving or reloading.
The gain needed for your target can push a transient above the sample-peak ceiling. This tool reduces the whole-file gain instead of limiting or compressing that transient, so loudness can remain below target. Absolute loudness gating and PCM16 quantization can also affect the measured result. The actual exported reading is shown.
No. It constrains stored sample peaks, not reconstructed inter-sample true peak. The tool does not measure true peak, loudness range, momentary loudness, or all requirements of a delivery specification. Measure your final delivery with suitable tools; later processing or lossy encoding can change its peaks and loudness.
No. One gain is applied to the complete recording and every channel. It does not automate individual words, compress dynamics, limit transients, or remove noise. A noise already in the source rises or falls with the wanted audio.
Choose a mono or stereo MP3/WAV up to 50 MiB, from 0.4 seconds to five minutes. Browser decoding is at 48 kHz; output is 48 kHz, 16-bit PCM WAV and must also fit within 50 MiB. Tags and artwork are not copied. A long stereo WAV can exceed the output limit even when its source MP3 fits.
No. It is an editable starting value, not a universal platform or broadcast requirement. Choose a target for your actual use and delivery specification. A trimmed excerpt has its own integrated loudness and cannot establish the loudness of a longer programme.
A silent or very quiet signal may have no complete analysis blocks above the meter's absolute gate. Without a measurable integrated loudness, the tool cannot calculate a meaningful target gain. It reports this rather than manufacturing a LUFS value or boosting silence.
After preparing your audio