Quick answer: in Sound Forge Pro 18 and the current 2026 release, open View → Channel Meters, right-click inside the meter, and enable Loudness Meters. Reset before playback, then run the complete file. M covers 400 ms, S covers three seconds, I accumulates Integrated loudness, LRA describes gated variation, and TP max catches reconstructed peaks. For delivery, judge I and TP max together.
The destination sets the target. EBU R128 uses -23 LUFS for programme loudness. The current Sound Forge help lists its ATSC preset at -24 LUFS/LKFS with TP max set to -2.0. Spotify adjusts Normal playback to -14 LUFS, but that is platform behaviour and mastering guidance, not a universal rule for music, podcasts, or video. These numbers are not interchangeable presets.
Use this guide for the live loudness meter and a complete playback check. For a static scan of a selection or file, use the Sound Forge Statistics guide. If the reading is already known and the file needs correction, use peak, RMS, and loudness normalization.
LUFS and Peak Answer Different Questions
A sample-peak meter reports the highest stored sample. That makes it good at finding obvious digital clipping, but poor at predicting how loud a complete programme feels. A sparse recording with one hard transient and a dense limited master can share a similar peak while sounding nothing alike.
LUFS describes programme loudness using frequency weighting and time integration. The current algorithm is ITU-R BS.1770-5, approved in November 2023. It is the measurement foundation under standards and recommendations such as EBU R128 and ATSC A/85.
True peak answers a third question. It estimates peaks that can appear between stored samples when digital audio is reconstructed or encoded. A WAV file may stay below 0 dBFS on an ordinary sample meter and still exceed a delivery ceiling in dBTP. That is why a loudness pass is not complete when Integrated LUFS alone lands on target.
How to Read M, S, I, LRA, and True Peak
Momentary loudness (M)
Momentary loudness uses a 400 ms window in EBU Mode. It moves quickly enough to expose a word, drum hit, edit, or sudden level jump, but it is not an instantaneous peak meter. Treat it as a locator. It tells you where to listen, not what number an entire file should hold.
Short-term loudness (S)
Short-term loudness uses a three-second window. It is usually the clearest meter for comparing a verse with a chorus, one narration sentence with the next, or adjoining sections recorded on different days. If the voice sounds steady but S swings several loudness units, check mic distance, clip gain, edit boundaries, or compressor release before reaching for a final limiter.
Integrated loudness (I)
Integrated loudness accumulates from the last reset across the material that actually plays. It uses gating so silence and very quiet passages do not distort the result like a simple arithmetic average. The reading is valid for the whole programme only after the whole programme has been measured. Replaying the loudest chorus three times, skipping the credits, or starting in the middle produces a number, but not the number the delivery specification asked for.
Loudness Range (LRA)
LRA is expressed in LU and summarizes the distribution of short-term loudness after absolute and relative gating. The EBU Tech 3342 specification deliberately prevents one gunshot, a long silence, or the noise floor from defining the entire range. It is therefore inaccurate to describe LRA as simply “loudest minus quietest.”
LRA is a diagnostic and editorial clue, not a universal quality score. A news read, an audiobook chapter, a classical recording, and a film mix should not share one mandatory LRA. The EBU itself cautions against using one maximum value for every programme. Ask whether the variation serves the material and the listening environment before compressing it.
True Peak (TP max)
TP max records the highest reconstructed peak during the pass. Read it in dBTP when making a delivery decision. If Integrated loudness passes but TP fails, lowering the final ceiling or revising the gain/dynamics chain is the fix. Renaming a meter preset or changing the LUFS target does not remove an inter-sample peak.
Open the Loudness Meter in Sound Forge on Windows
- Open the finished or nearly finished audio file.
- Choose View → Channel Meters. In the Pro 18/current-family interface, the loudness display is integrated into the Channel Meters window.
- Right-click inside the meter area and enable the Loudness Meters view or choose a Loudness Meter preset.
- Open Options → Preferences → Status.
- At Loudness Configuration, select EBU R128 or ATSC A/85 according to the delivery sheet.
- Right-click the loudness display and choose Reset before the full pass.
- Return to the beginning and play the required programme without skipping sections.
The current Boris FX shortcut list assigns Alt+6 to the Loudness Meters window. If the shortcut opens nothing, choose View → Channel Meters and enable the layout from the meter’s context menu.
Older MAGIX-era builds can use different labels or panel layouts. This guide follows the current Boris FX Windows line; the Sound Forge free-trial guide covers the current installer and version path.
The Sound Forge Display Trap: LU Is Not LUFS
Sound Forge can show loudness relative to the selected target in LU or as an absolute value in LUFS. In relative mode, 0 LU means “at the target.” Under EBU R128 that corresponds to -23 LUFS; under the Sound Forge ATSC preset it corresponds to -24 LUFS/LKFS. A reading near 0 is not 0 dBFS and it is not clipping.
For screenshots, logs, and client handoff, absolute LUFS is harder to misread. Right-click the meter, inspect the range/display options, and enable the absolute display when you need the actual full-scale value. Use relative LU when riding a programme around a known target.
EBU R128, ATSC A/85, and Streaming Targets
The table below separates standards from platform behaviour. It is not a list of interchangeable mastering presets.
| Destination | Loudness reference | True peak | What the number means |
|---|---|---|---|
| EBU R128 v5.0 | -23 LUFS programme loudness | -1 dBTP permitted maximum in the core recommendation | Broadcast programme normalization; use the broadcaster’s exact delivery sheet for tolerances and supplements. |
| Sound Forge ATSC preset | -24 LUFS/LKFS | TP max -2.0 in the current help | An application preset. Use the delivery sheet when its target or ceiling differs. |
| ATSC A/85:2026-07 | -24 LKFS by default without metadata; long form is dialogue-anchored, short form uses full-program measurement | -2 dBTP | Current television and streaming guidance. For streaming services without another agreement, Annex M recommends one service target from -23 to -27 LKFS. |
| Spotify Normal | -14 LUFS playback normalization | Spotify recommends at most -1 dBTP at -14 LUFS, or below -2 dBTP for louder masters | Playback behaviour and platform mastering advice, not a broadcast acceptance rule or a mandate for every music master. |
| Podcast, audiobook, social video | Publisher-specific | Publisher-specific | Use the host, network, distributor, or client specification. “The internet uses -14 LUFS” is not a delivery document. |
EBU R128 version 5.0 remains centred on -23 LUFS, LRA, and Maximum True Peak. The current ATSC A/85:2026-07 is more nuanced than the old one-line “-24 LKFS” summary. Its Annex M quick reference distinguishes dialogue-anchored long form, full-program short form, metadata workflows, and streaming services.
LKFS and LUFS are numerically synonymous labels here. A target of -24 LKFS is not one unit different merely because another meter writes LUFS.
A Full-Programme LUFS Measurement You Can Trust
- Read the delivery specification. Record the loudness target, tolerance, true-peak ceiling, measurement anchor, channel layout, codec, and file format.
- Choose the matching mode. Use EBU R128 or ATSC A/85 in Sound Forge. If the client sheet differs from the preset, the sheet wins.
- Check channel interpretation. Six-channel files can use surround processing in Preferences. Don’t measure an incorrectly ordered multichannel file as if every channel carried equal weight.
- Reset the meter. Clear Integrated, LRA, and held true-peak data from previews.
- Play the complete required programme. Include the opening, pauses, credits, and tail unless the delivery specification excludes them.
- Read I and TP first. These values usually decide pass or fail. Use M, S, the history curve, and LRA to locate the cause.
- Correct the smallest real problem. A gain move may be enough. If peaks block the loudness target, revise dynamics or limiting instead of normalizing repeatedly.
- Render the required file. Use the final codec and settings, not a convenient temporary format.
- Measure the rendered file again. Encoders can change peak behaviour. The file you deliver is the file that must pass.
This sequence is slower than reading a number from the loudest section, but it removes the two common unknowns: stale Integrated data and codec-induced peak changes.
Diagnose the Failure Before You Process
| Meter result | Likely issue | First move |
|---|---|---|
| I too low, TP has headroom | The whole file is simply quiet | Raise gain by the measured difference, then run another full pass. |
| I too low, TP already at the ceiling | Transients or dynamics block a clean gain increase | Control the specific peaks or revise compression/limiting before adding gain. |
| I passes, TP fails | Inter-sample or codec-sensitive peaks exceed the ceiling | Lower the final true-peak ceiling or use a suitable limiter; verify the render. |
| I passes, speech still jumps | Whole-program average hides section inconsistency | Use S and the history trace to find edits, mic changes, or clip-gain problems. |
| LRA looks wide | Could be creative dynamics or inconsistent production | Listen in context; do not compress merely to reduce the number. |
| Two meters disagree | Different reset periods, channel layouts, gating, or algorithm versions | Match start/stop points, mode, channels, and BS.1770 compliance before comparing. |
If the file needs a final dynamics stage, Wave Hammer can combine compression and limiting, but it should follow diagnosis rather than replace it. For a fuller single-file chain, use the Sound Forge mastering walkthrough.
Generate a Loudness Log: Measure at the Right Point
Sound Forge can document a pass with Tools → Generate Loudness Log; the current help also uses the label Create Loudness Log on the same page. With no selection, it analyzes the whole file; with a selection, it reports that range. The log records the file, format, metering mode, and loudness values and is useful when a broadcaster or client expects evidence with the master.
There is an important codec detail in the Sound Forge manual: a loudness log requested during Save As is calculated after the plug-in chain but before codec encoding. For WAV delivery, that may be exactly what you need. For MP3, AAC, or another compressed deliverable, open the saved file and generate the log afterward so the measurement includes any peak change introduced by encoding.
For MP3 settings and verification, follow the Sound Forge MP3 export guide. Don’t approve the encoded file from the pre-codec meter alone: the encoder is the last thing that touches your peaks.
Meter, Normalize, and Limiter Are Three Different Jobs
The Loudness Meter measures; it does not alter audio. Process → Normalize can make a calculated gain or loudness-normalization move. A limiter controls peaks and changes dynamics when gain alone cannot reach the target safely.
Suppose a file measures -25 LUFS and -5 dBTP against a -23 LUFS / -1 dBTP delivery. A roughly 2 LU gain increase has enough headroom and may solve the problem without compression. If the same file measures -25 LUFS and -1.2 dBTP, adding 2 dB will violate the peak limit. That is not a “normalize harder” problem; the crest factor or specific transients need attention.
Sound Forge’s loudness normalization can be useful for repeatable files, but always remeasure. On a folder job, build and test one representative file before applying the process through the Sound Forge Batch Converter.
Why -14 LUFS Is Not a Universal Mastering Target
Spotify currently states that its Normal setting adjusts playback to -14 LUFS under ITU 1770. It measures uploads, then applies gain during playback rather than rewriting the uploaded master. Spotify also notes that behaviour can differ on the web player and third-party devices.
That does not mean every music master should be forced to exactly -14 LUFS. Genre, crest factor, distortion, album continuity, and the sound of the limiter still matter. A louder master may be turned down; a quieter, highly dynamic master may not be raised all the way if true-peak headroom runs out. The correct editorial decision is to make the master sound intentional, keep enough encoding headroom, and understand the destination’s playback behaviour.
Broadcast is different because the number can be an acceptance condition. “Spotify turns tracks down” is not an argument for sending a -14 LUFS programme to a broadcaster asking for EBU R128 at -23 LUFS.
Common LUFS Meter Mistakes in Sound Forge
- Not resetting: the Integrated result still contains previews.
- Measuring only the loudest passage: the result is not full-program loudness.
- Reading relative LU as absolute LUFS: 0 LU means target, not full scale.
- Watching sample peak instead of TP max: the file can fail after reconstruction or encoding.
- Treating M or S as a fixed master target: these windows diagnose passages.
- Calling LRA loudest minus quietest: the standardized measure uses gating and a distribution.
- Using one streaming number everywhere: platform playback, broadcast exchange, and a client delivery sheet are different contexts.
- Logging before the codec only: the encoded deliverable may have a different true peak.
- Normalizing repeatedly: gain changes do not repair uncontrolled dynamics.
- Ignoring the anchor: current ATSC long form can be dialogue-anchored rather than a simple full-mix average.
LUFS Meter FAQ
Where is the LUFS meter in Sound Forge?
Open View → Channel Meters, then use the meter’s right-click menu to show the Loudness Meters layout. In the Pro 18/current-family interface, Loudness Configuration is under Options → Preferences → Status. Alt+6 shows or focuses the loudness window.
What is Integrated LUFS?
Integrated LUFS is the gated loudness value accumulated across the complete measurement period. Reset the meter and play the whole required programme before using I as a delivery result.
What is the difference between LUFS and LKFS?
For BS.1770-based programme loudness, LUFS and LKFS are numerically synonymous labels. -24 LKFS and -24 LUFS describe the same loudness value.
Should I choose EBU R128 or ATSC A/85?
Choose the standard named by the broadcaster, platform, or client. EBU R128 uses -23 LUFS. The current Sound Forge help lists its ATSC preset at -24 LUFS/LKFS with TP max set to -2.0, while ATSC A/85:2026-07 expresses the maximum as -2 dBTP and adds rules for dialogue anchoring, metadata, short form, and streaming.
What is a good LRA number?
There is no universal good LRA. It depends on genre, programme type, and listening environment. Use LRA to spot inappropriate variation, then listen before changing dynamics.
Why is True Peak higher than sample peak?
True Peak estimates the reconstructed waveform between stored samples. Inter-sample peaks can exceed the highest sample value and may rise further after lossy encoding.
Can Sound Forge create a loudness report?
Yes. Use Tools → Generate Loudness Log (also called Create Loudness Log in the current help) for a file or selection. For compressed delivery, generate the log from the saved encoded file because save-time logging occurs before the codec.
Should I normalize every file to -14 LUFS?
No. -14 LUFS is Spotify’s Normal playback reference and mastering guidance, not a universal delivery target. Follow the destination specification and preserve sound quality.
Verify the File You Will Deliver
Choose the destination first. In Sound Forge, select the matching loudness mode, reset the meter, play the complete programme, and read Integrated loudness beside TP max. Use M, S, LRA, and the history trace to explain the result, not to invent a second target. Then render and measure the actual delivery file. The file you send, not the pre-render meter pass, is the result that must meet the specification.