To normalize Broadcast WAV files without losing their metadata, use a tool that copies the bext, iXML, LIST and cue chunks into the corrected file instead of writing a fresh plain WAV, then check the result with a metadata viewer. WavSpec does this on Windows: it carries those chunks over in their original order, copies unknown chunks as they are, and when a spec asks for it writes the verified loudness into the bext chunk using the EBU Tech 3285 version 2 fields. The corrected files go to a separate delivery folder, and the originals are never overwritten.
Losing metadata is easy to miss. The file plays, the loudness is right, and only later does an editor find the timecode reference is gone and the clip lands at zero on the timeline.
Correct loudness on Broadcast WAV files and keep their bext, iXML and cue chunks.
Get it from Microsoft
What metadata does a Broadcast WAV file carry?
A Broadcast Wave Format (BWF) file is a WAV file with extra chunks. The EBU defines the core one, the broadcast audio extension or bext chunk, in EBU Tech 3285. Other chunks often travel with it.
| Chunk | What it typically holds | Why losing it hurts |
|---|---|---|
| bext | Description, Originator, OriginatorReference, OriginationDate, OriginationTime, TimeReference, UMID, loudness values (version 2), CodingHistory | TimeReference places the file on a timeline; the description and originator identify it |
| iXML | Production metadata many field recorders write, such as scene, take and track names | Editors and conform tools use it to find the right take |
| LIST (INFO) | Basic tags such as title and comments | Library and catalogue tools read it |
| cue | Markers inside the file | Edit points and notes disappear |
The bext field list is from EBU Tech 3285 version 2, checked on 17 October 2026. The other descriptions are general; check the specific fields your recorder and editor use.
TimeReference is the field that hurts most when it goes. Tech 3285 defines it as the first sample count since midnight, which is how an editor places a sound file at its original timecode.
How does bext version 2 store loudness?
Version 2 of the bext chunk, which the EBU says replaced version 1 in May 2011, uses 10 of the reserved bytes for five loudness fields:
| bext v2 field | Meaning | Stored as |
|---|---|---|
| LoudnessValue | Integrated loudness of the file in LUFS | 16-bit integer, value multiplied by 100 |
| LoudnessRange | Loudness range in LU | 16-bit integer, value multiplied by 100 |
| MaxTruePeakLevel | Maximum true peak in dBTP | 16-bit integer, value multiplied by 100 |
| MaxMomentaryLoudness | Highest momentary loudness in LUFS | 16-bit integer, value multiplied by 100 |
| MaxShortTermLoudness | Highest short-term loudness in LUFS | 16-bit integer, value multiplied by 100 |
Source: EBU Tech 3285 version 2, checked on 17 October 2026.
So a file at -23.00 LUFS stores -2300 in LoudnessValue. The specification says a parameter that is not used is set to 7FFFh, so readers know to ignore it. It also warns that version 0 and version 1 software ignores the loudness fields, so older tools can lose them unless special precautions are taken.
For loudness terms, see our guide to LUFS, LKFS, dBTP and LRA.
Why does normalizing sometimes strip metadata?
Changing gain means rewriting the audio data. Some tools then write a new WAV file containing only the format and data chunks, because that is the minimum a WAV needs. Every other chunk, including bext and iXML, is dropped. Resampling or converting bit depth makes it more likely, because the file is rebuilt from scratch.
There is a second trap: a tool that keeps the bext chunk but leaves the old loudness values in it. After correction those values describe the file as it was, not as it is. A downstream system that trusts bext loudness would then read the wrong number.
How do I check whether a tool kept the metadata?
Test before you trust a workflow:
- Pick a test file with a known bext description, a TimeReference and some iXML.
- Inspect it first. BWF MetaEdit, a free tool initially funded by the Library of Congress and FADGI and developed by MediaArea, shows and exports bext and INFO metadata for Windows, macOS and Linux.
- Run the normalization you plan to use.
- Inspect the output. Compare the description, originator, origination date and time, TimeReference, and whether iXML is still there.
- Drop the output into your editor and check it lands at the original timecode.
How does WavSpec keep the chunks?
WavSpec treats metadata as part of the delivery:
- Originals stay untouched. Source files are opened read-only, and corrected copies go to a separate delivery folder.
- Chunks are carried over in order. The bext, iXML, LIST and cue chunks go into the delivered file in their original order, and unknown chunks are copied as they are, so time references and scene and take data survive.
- Loudness is written after the fix. When the spec asks for it, WavSpec writes the verified loudness values into bext using the Tech 3285 version 2 fields, measured from the delivered file, not copied from the source.
- Large files stay large. WavSpec reads RF64 as well as WAV and BWF.
- The report records it. The PDF, HTML, CSV and JSON report lists before and after values per criterion and the SHA-256 of every delivered and source file.
The built-in EBU R 128 broadcast preset, for example, delivers each file at -23 LUFS within tolerance, under -1 dBTP, at 48 kHz, with its values in bext. More on that workflow in normalize audio to EBU R 128.
Run a test file through WavSpec in the 15-day free trial and inspect its bext afterwards.
Get it from Microsoft
Should I normalize the audio or just write loudness metadata?
For delivery to a spec, change the audio. A broadcaster’s QC checks the measured signal, and a file at -19 LUFS with -23 written in its metadata is still a -19 LUFS file. EBU Tech 3343, the guidelines that accompany R 128, recommends normalizing the signal itself and treats metadata-based normalization as an alternative. Writing correct loudness values into bext after the correction gives downstream systems an accurate description without asking them to trust it.
When is another tool the better pick?
- You need to edit metadata by hand. BWF MetaEdit edits and validates bext fields and is free. WavSpec writes loudness values; it is not a general metadata editor.
- You are still in the session. Your DAW or field recorder software may keep iXML through its own export, and that may be all you need.
- You need macOS or Linux. WavSpec is Windows 11 only.
Frequently asked questions
Does WavSpec keep TimeReference after resampling?
WavSpec carries the bext chunk into the delivered file. Because TimeReference is a sample count, check in your editor that the delivered file lands at the expected timecode after any sample-rate change, as you would with any tool.
Does WavSpec write loudness into bext for every spec?
It writes the verified loudness into bext v2 fields when the spec asks for it. Built-in broadcast specs such as EBU R 128 do.
Does WavSpec handle MP3 metadata the same way?
The bext, iXML, LIST and cue chunks belong to WAV-family files. MP3 deliveries, such as audiobook chapters, are encoded as CBR MP3.
Are my originals changed?
No. Originals are opened read-only and never overwritten.
Does WavSpec need an internet connection?
No. It opens no network connection, has no account and collects no telemetry.
Try WavSpec
WavSpec is a one-time purchase in the Microsoft Store, with a 15-day free trial. Run a production WAV through the EBU R 128 preset and inspect the delivered file’s bext and iXML. For the full delivery list, see our audio delivery QC checklist, or go to the WavSpec product page.
Fix the loudness and keep everything else the file carries.
Get it from Microsoft
EBU is a trademark of the European Broadcasting Union. Microsoft and Windows are trademarks of the Microsoft group of companies. Other names are trademarks of their owners. WavSpec is not affiliated with or endorsed by any of them.
Last updated 17 October 2026. Published by Automata Labs, a software company in Canada.