What's New in DiffALL
What we shipped recently, and what it lets you do. Newest first.
-
Free audio comparisons now cover 2 minutes
The free plan now compares 2 minutes of audio, the same as Pro, with or without an account.
- A recording longer than 2 minutes, up to 3, is accepted on every plan: choose where the 2 minutes start, or leave it at 0:00. The result says which part was compared.
- Pro still compares whole videos end to end and takes larger files.
-
Every quality check now shows its score — and where the problem is
A quality report used to name a problem. It shows it now: each check carries a 0-100 score on its own scale, and where the finding is about a place or a moment, the report points at it.
- Each check gets its own number, never one for the file. An average would hide the one thing that is wrong, so the headline is still the weakest check — the number just says how weak.
- Images: clipped pixels painted on your own photo, the strongest JPEG blocking, the flat areas where noise shows, and on a soft shot the sharpest regions, so you can see that even those are soft. Video: the worst moment's frame, marked.
- New: a video that freezes mid-motion, and audio that drops to silence and straight back, are listed with their timestamps. Neither can be told apart from a deliberate hold or a hard edit, so both are offered as moments worth a look rather than a mark against the file.
- Two bugs fixed on the way: clipping in one stereo channel was being missed entirely, and a clipped moment in a segment was shown at the wrong time.
- Video reports also read what the container itself says, which costs nothing and no other check can tell you: frames that never arrived and where, a variable frame rate, audio that starts off the picture, a rotation flag, non-square pixels, 10-bit, missing colour tags. Each one says what it means for you.
- Images and audio do the same. A photo report can now tell you the JPEG quality it was saved at, that it carries a rotation flag or the location it was taken, and — beside a noise or softness finding — the ISO or shutter speed that caused it. An audio report says whether the bitrate is constant or an average.
-
Two files, or one — the home page asks first
DiffALL does two things now, and the home page says so before it asks for anything. Choose Compare Any Two Files to see exactly what changed between a before and an after, or Measure Any File's Quality to find what is wrong with a single video, image or audio file — no second file needed.
- Quality checks are free on every plan. A paid plan reads more of a long clip — up to 3 minutes of video or audio instead of the first 10 or 60 seconds.
- The quality check on the home page takes any of the three, with a sample photo, clip and audio file to try it on. The audio page's own sample button was sending its MP3 as a picture and failing; that is fixed.
- The API does both too: POST one file to /api/v1/quality and get the same report the site shows, one credit per call like a comparison.
-
Audio Quality Check — and what your file is really carrying
The quality check now takes audio, which completes the set: video, images and now sound. Upload one audio file and it measures loudness in LUFS against the level every streaming platform normalises to, finds clipped samples, reads how much of the frequency range is actually there, and says whether a stereo file carries two channels or one twice.
- It catches a file that was a low-bitrate MP3 earlier in its life. A 64kbps file re-exported at 320kbps says 320kbps in every byte of metadata and still has 64kbps of content in it — measured, that reads as an Excellent bitrate beside a frequency range that stops at 10.9kHz.
- Loudness is ITU-R BS.1770 with the filters derived at your file's own sample rate, not the 48kHz constants most loudness code copies and then applies to 44.1kHz audio.
- Clipping counts the longest flat run, not just samples at full scale — so a master limited to -1dB is not called clipped, and a waveform with its top cut off is.
- There is deliberately no noise or hiss check. Nothing separates hiss from quiet content without a reference, and a check that calls a field audio file noisy is worse than no check at all.
-
Video Quality Check — including whether the frame rate is real
The quality check now takes video as well. Upload one clip and it samples it second by second: bitrate against the recommended upload rate for that size, focus, noise, compression, clipping, and whether consecutive frames are actually different.
- Bitrate is measured as bits per pixel per frame, so one number means the same thing for 720p30 and 4K60, and it is compared against the published recommended upload rates rather than against an opinion.
- It catches a frame rate that is not real — a file can claim 60fps and carry 30 distinct pictures, which is invisible until you scrub.
- The band is the MEDIAN second, not the worst: one soft second in a whip pan is motion, not a defect. The worst moment is named beside it, as a place to look.
- Free checks the first 10 seconds, Pro the whole upload up to 3 minutes — and the report always says exactly how many seconds it read.
-
Image Quality Check — one file, no before-and-after
Every other tool here needs two files and needs you to know which is which. This one takes a single image and tells you what is wrong with it: how sharply focused it is, how much noise is in it, how much JPEG block structure it carries, and how much of it is blown out or crushed to black. Each one gets a plain verdict and, where there is something to do about it, what to do.
- No overall score out of 100, on purpose. You get the WEAKEST of the checks named at the top — an image that is perfectly exposed, noise-free and completely out of focus is not 75% good, and an average would say it was.
- It tells you when it cannot tell. A plain gradient gets no focus verdict because there is nothing in it to focus on; a flat-fill screenshot gets no compression verdict because that check needs photographic detail to read. Silence beats a confident wrong answer about your work.
- Free and complete, no sign-up, and the image is deleted the moment the check finishes.
- Got the original too? The report links straight through to the image comparison, which shows exactly what an export or an edit changed.
-
Voice Comparison — is it the same person speaking?
Upload two recordings of someone talking and find out whether it is the same person. It compares the characteristics of a voice rather than the words, so the two clips can be in different languages, recorded on different devices, and say completely different things.
- Three answers — likely the same speaker, likely different, or not conclusive.
- It refuses to guess: under three seconds of speech it tells you so instead of returning a number you cannot trust.
- Two one-click samples on the page, one of each answer, so you can see it work before uploading anything.
-
Longer clips, and the tool on every page
Video uploads now go up to three minutes on every plan, including free — how much gets analysed still depends on the plan, but the upload no longer stops you at the door. Audio limits went up too.
- Video: up to 3 minutes on any plan.
- Audio: 90 seconds free, 3 minutes on Pro — compared end to end. (Updated 2026-09-14: the Pro audio limit is now 2 minutes. The original figure was set against a memory measurement that counted shared library pages the server is only charged for once, so 3 minutes never actually fit. Updated 2026-10-04: the free plan is now 2 minutes as well.)
- Every comparison page now carries the tool itself, so you can compare from the page you landed on instead of being sent back to the homepage.
-
A credits pack, and a bigger free allowance
Not everyone wants a subscription for something they need a few times. There is now a one-time pack, and the free allowances went up at the same time.
- Credits pack: $5 once for 50 comparisons, with Pro-level file and length limits. No subscription, nothing to cancel.
- Signed in and free: 10 comparisons a day, up from 5.
- No account at all: 5 comparisons a day.
-
Free video gets a preview instead of a wall
A free video comparison over the length limit used to be refused outright. Now it runs on the opening seconds and gives you a real score and heatmap for that stretch, so you can see what the tool actually does before deciding whether it is worth paying for the rest.
Want the detail behind the scores? Read how the comparisons work.