Your audio sounds worse after uploading.
The platform re-encoded it, and Vimeo publishes how far down.

This is the rare quality question that does not need a measurement. Vimeo lists a maximum delivered audio bitrate for every rendition on its ladder, and the numbers are lower than the file you handed it. The 240p rendition is capped at 64 Kbps and delivered as HE-AAC V2 or Opus. The 360p rendition is capped at 129 Kbps. From 540p up through 4K the maximum is 256 Kbps, delivered as AAC or Opus. The HDR 2K and 4K renditions top out at 265 Kbps. Every conversion is resampled to 48 kHz.

Now set that beside the other page. Vimeo’s compression guidelines ask you to upload audio as AAC-LC at 320 kb/s CBR at 48 kHz. A platform that asks for 320 and delivers a maximum of 256 is re-encoding your mix downward, and it is telling you so in its own documentation. Nobody has to take a measurement or trust a benchmark. Two help-center pages, two numbers, one gap.

One qualifier that matters for honesty: Vimeo says the listed bitrates are a maximum, and that a video could transcode lower depending on the complexity of the image. So the correct reading is “up to 256 Kbps”, never “always 256 Kbps”. The ceiling is the best case.

It is worth understanding why the audio gets rebuilt at all, because the reason is not carelessness. A streaming platform does not store your file and send it out; it builds a set of renditions at different resolutions and bitrates so a player can switch between them as a connection moves. Audio travels inside those renditions. Once the platform has decided to rebuild the picture for a 240p rung that has to survive a bad mobile connection, the audio riding along with it gets a budget to match, and 64 Kbps is that budget. The audio ceiling is not a separate decision about your mix. It is a consequence of the ladder existing.

The video column is worth a glance even on an audio page, because it tells you how much the ladder is doing. Vimeo markets a ceiling of 4K Ultra HD, and that ceiling arrives at up to 22 Mbps. The rendition that carries 256 Kbps of audio is the same rendition carrying 22 Mbps of picture. The one that carries 64 Kbps of audio is carrying 0.3 Mbps of picture. The audio ladder and the video ladder fall together, which is why a viewer who complains that the picture looks soft usually has a thin-sounding mix as well.

The sample rate deserves a line of its own. Vimeo states that the audio sample rate for all conversions is 48 kHz. For most film and television work that changes nothing, because 48 kHz is already the delivery standard. It does matter if your mix came out of a music session at 44.1 kHz, or at 96 kHz from a sound-design session: that track is resampled on the way through, and a resample is a processing step you did not sign off on. Print the delivery at 48 kHz yourself and the conversion becomes a no-op instead of a surprise.

Vimeo gates the top of the ladder on the uploaded dimensions: a 4K rendition exists only if the larger dimension is at least 3,648 pixels or the smaller is at least 2,052; 2K needs 2,432 or 1,368; 1080p60 needs 1,824 or 1,026. Below those thresholds the rendition is never built, so the audio bitrate attached to it is never available either. Above them it exists, but the player still has to choose it.

A mix that survives 256 Kbps can come apart at 64. That is not a subtle difference, and it is not distributed evenly across the mix. Some material is nearly immune and some falls over first.

This is also why the note that comes back from a client so often makes no sense to the person who mixed it. You approved the film on a fast connection, on the top rendition, at 256 Kbps. Your client opened the link on a laptop tethered to a phone, the player settled on a low rung, and the track they heard was a different encode of a different bitrate. Two people described the same file honestly and disagreed, because they were not listening to the same file. Nothing in the interface tells either of them which rendition they got.

Here, the player streams the exact bytes you uploaded. The audio track plays as exported: no AAC or Opus re-encode, no resample away from your sample rate, no loudness pass, because nothing is transcoded at any point in the path. There is one file, the one you made, and every viewer gets it. Uploads are MP4 or MOV carrying H.264, HEVC or AV1, up to 50 GB per file. What you approved in the mix room is what arrives.

The practical effect on a sound approval is that the conversation stops being about the platform. When a composer, a sound designer and a director all open the same link and hear the same bytes, a note about a reverb tail is a note about the mix, and it can be answered in the review room against a timecode instead of being argued about for a week. Nobody has to ask which rendition the other person got, because there is only one. That is a smaller claim than it sounds: it does not make the mix better, it just removes a variable that had no business being in the room.

The same reasoning applies to the picture, and the mechanism is identical: one file, delivered whole, instead of a set of rebuilt approximations. If the visual side is the part that bothers you, the longer version of that argument is in video hosting that does not compress and in why your video looks worse after upload.

If your delivery target is a broad public audience watching on phones, on mobile data, in transit, on hardware you will never see, then a ladder that includes a 64 Kbps rendition is what keeps the video playing at all. A thin-sounding track that plays is worth more than a perfect track that stalls. That is the whole design intent of adaptive delivery, and it works.

There is no ladder here, and that trade runs both ways. A viewer on a poor connection buffers instead of silently dropping to a low rendition, which is the right behaviour for a client reviewing a grade and the wrong behaviour for a marketing video aimed at everyone. Playback also depends on the viewer’s machine having a hardware decoder for the codec you uploaded, which is a safe assumption for a post house and a shakier one for the general public.

So the split is straightforward. Public reach, phones and unknown devices: use a platform with a ladder, accept the audio ceiling, and mix so the important material survives 64 Kbps. Client review, grade approval, sound-design approval, festival and portfolio delivery, anywhere the point is that someone hears the work as it was finished: send the file itself. Vimeo’s player, marketing tools and OTT products have no equivalent here, and if those are what you are paying for, this page is not asking you to leave.