Before you send a WAV to another producer, check five things: it is a real uncompressed WAV, its sample rate and bit depth are what you meant, nothing in it clips, it has a little headroom, and it is trimmed to the length you promised. Most of this takes under a minute, and a file that fails any one of them is the kind a collaborator quietly stops answering.
The quickest way is to drop the file into the WAV file checker, which reads the header and every sample in your browser and never uploads anything. This guide explains what each line of its result means, how to check the same things in a DAW or your operating system, and what to do when something is off.
Why check at all
A file looks the same in a folder whether it is right or wrong. The mistakes that cost a collaboration are the invisible ones: a bounce that clipped on the master, a 48 kHz file you thought was 44.1, a loop with half a second of silence at the end, an MP3 someone renamed to .wav. None of them announces itself until a stranger has dropped the file into a session and it sounds wrong.
Checking also settles an argument before it starts. If you can say "24-bit, 48 kHz, peaks at -3 dB", nobody has to guess what they were sent.
Is it actually a WAV?
A WAV is a container, and the container can hold things that are not plain PCM audio. The file extension tells you nothing; the header does. A real send-ready file is uncompressed PCM (integer or 32-bit float). Compressed formats such as ADPCM, and files that are really MP3 or AAC with a .wav name, will open in some tools and fail in others.
The checker reads the format tag and reports it. It also understands the "extensible" header that many DAWs write for 24-bit and multichannel files, so a file is not flagged for that alone. If it says the file is not a WAV at all, export again from the source. Do not rename your way out of it. Sample rate and bit depth explains why an MP3 is the wrong thing to build on.
Sample rate and bit depth
Look for the rate you worked at, almost always 44.1 or 48 kHz, and 24-bit (or 32-bit float for stems you want left untouched). 16-bit is fine for a finished loop but leaves less room if the collaborator turns it up.
The mistake to catch is a mismatch with your project. If you worked at 48 kHz and the file says 44.1, something converted it on the way out, and the length may no longer land exactly on a bar. Do not resample upwards to make a file look more professional; a converter cannot add detail that was never recorded. The reasoning and the file-size arithmetic are in the sample rate and bit depth guide, so it is not repeated here.
Mono or stereo
The channel count is worth a glance. A single bass line or vocal exported as two identical channels doubles the file for nothing. A pad exported as mono has lost its width. If the checker shows 2 channels and you meant a mono part, check your export dialog for a "mono" or "split channels" option.
Peak level, clipping and headroom
This is where the file itself needs reading, because you cannot see clipping in a file name.
- Peak is the loudest single sample, in dBFS, where 0 dBFS is the top of the scale. A peak of -3 dBFS means the loudest moment sits 3 dB below the ceiling.
- Clipping is what happens when the signal tried to go past 0 dBFS and was flattened at the ceiling. In the file it appears as a run of samples all sitting at full scale. A single sample touching the ceiling is usually just a loud peak. A run of them is a flat-topped wave, and once it is baked into a 16- or 24-bit file the lost detail cannot be recovered by turning anything down.
- Headroom is the gap between your peak and 0 dBFS. Aim for a peak of about -1 dBFS or lower. A collaborator will add gain, an effect or a layer, and a file that already touches the ceiling has no room for any of it.
The checker reports the peak, a rough RMS (an average level, useful for comparing two bounces), and flags real clipping separately from a loud peak. If it finds clipping, the fix is upstream: lower the master fader or the loudest channel and bounce again. Normalising the clipped file only makes the damage louder. Should you mix loops? covers how much headroom to leave and why you should not put a limiter on a loop you are handing over.
32-bit float files can hold levels above 0 dBFS without damage, so a float peak over zero is not yet clipping. It will clip the moment it is converted to an integer format, so pull it down before you send.
Length, trimming and whole bars
Check the duration against what you promised and against the platform's rules. A ColabLoop loop is between 10 and 60 seconds and up to 30 MB, and the checker's list says which side of those limits you are on.
Then check the ends. Dead silence at the start or a long tail of nothing at the end is the usual trim error. For a loop, the real test is whether its length is a whole number of bars at your tempo: type the BPM into the checker and it tells you how many milliseconds the file is from the nearest bar line. A few milliseconds is a rounding error; tens or hundreds means the loop will drift when it repeats. How to make seamless loops walks through exact bar boundaries and the tail-wrap method.
Metadata a file may be carrying
A WAV can hold more than audio, in extra chunks after the format information. Three are worth knowing:
- ACID tempo and root note. Loop-oriented tools write the tempo and sometimes the root note here, and samplers use it to time-stretch automatically. If it is wrong, a collaborator's sampler will stretch your loop to the wrong speed without telling them.
- Sampler loop points (smpl). Marks where a loop starts and ends. Wrong points make a sampler loop the wrong region.
- INFO tags, broadcast metadata and ID3. Text such as the software that wrote the file, a title, or an originator. Harmless, but you may not want your full name or a project name travelling inside the file.
The checker lists every chunk it finds, plus the tempo and loop points where they exist. Most WAVs carry none of this, and as how to name loop files says, you cannot rely on any tag surviving the trip. The filename is what reliably travels, so put the key and tempo there as well.
How to check in your DAW or operating system
You can check most of this without a separate tool.
- Windows. Right-click the file, open Properties, then Details. It shows the audio sample rate, bit depth and channels, and the length.
- macOS. Select the file in Finder and press Command-I, or open it in QuickTime and press Command-I. Sample rate and duration appear under More Info.
- Any DAW. Drop the file onto an audio track. Most show the sample rate and bit depth in the file or clip properties, and the meter shows the peak while it plays. If the clip does not line up with the grid at your project tempo, the length is off.
- Metering. A true-peak or sample-peak meter on the final bounce is the standard way to see the ceiling. Play the whole loop through it once and read the maximum, not just what you see during the loudest bar.
None of these show clipping across the whole file or the chunk list, which is why the WAV file checker reads every sample.
What to fix
- Not PCM, or really an MP3: export a fresh WAV from the project.
- Wrong rate or depth: export again at your project's rate and 24-bit; do not convert upwards.
- Clipping: lower the loudest source, not the finished file, and re-bounce.
- No headroom: drop the master so the peak sits around -1 dBFS or lower.
- Too long or too short: trim to a whole number of bars, and keep it within 10 to 60 seconds for ColabLoop.
- Off the bar line: re-export from an exact bar start to an exact bar end.
- Wrong tempo in the file's metadata: re-export, or ask your DAW to write no tempo tag rather than a wrong one.
Fix it, bounce it again and re-check it. When the file is clean, send it with a clear name, and if you want other producers to build on it, you can publish it on ColabLoop. Nothing is sold: they ask, you decide, and credit is agreed before anyone starts.