Copy link to clipboard
Copied
OK, I've been going to town on researching this matter all day.
Audio for film has always, traditionally, been at 48k - whilst CD's and digital music, generally, at 44.1k.
Now... YouTube used to only allow 44.1k audio on upload - as did Vimeo... Of course, they sorted this out and started to accept 48k. All fine so far.
However... whilst YouTube and Vimeo accept 48k - they apparently down sample to 44.1k for their videos. And so this brings up a very important point!!!!
I know of people (and I was about to do the same) that when editing music videos; always receive the audio master of the song, as a 44.1k file (usually 24 Bit 44.1k)... some people have been upsampling to 48k, as that what they assumed was the video standard... only to have the video service downsample to 44.1k for streaming! The point being that: two needles sample rate conversions have taken place!!!!! Resulting in poorer audio.
Is this a widely known fact? As I have found little to nothing about this online and I feel it needs making clear. For anyone editing music videos, especially.
Have I got any of this wrong - or does anyone have anything to add to this??
Copy link to clipboard
Copied
The audio settings for video files has always seemed rather ... messy. Different formats & codecs do slightly different things, and so even in the encoding you may think you're getting one thing, then ... check say through MediaInfo, and find that actually the created isn't exactly what you thought.
Now ... throw in uploading and then downloading & playback on any one computer. Those can all be throwing a brick in the process.
Coming into video post from somewhat of an audiophile background, I wanted everything at LEAST at 48k, if not 96! And found it odd that most software seemed to assume 44.1k. But in talking with the others in aisles at NAB the last few years, so much of the hardware in broadcast, cable, and satellite use is built around 44.1k, that it auto-changes anything coming in to that setting. Trying to get anything to actually stay in 48k through export, uploading, storage on the service, downloading from their service, and travel over the ISP's system into a specific computer is probably an exercise in total futility.
So ... 44.1k is pretty much the highest sampling rate you can guarantee ... and actually get delivered especially via the web.
It is what it is ... trite and annoying phrase, but ... accurate here.
Neil
Copy link to clipboard
Copied
Thank you. I just spoke with a Vimeo support person and they said they stream audio at 48k - which I think they are wrong on... I'm sure it is 44.1k. They let you upload at 48 but I don't believe it stays that way.
My source file is 44.1 anyway! I'm fine either way - I just want to minimise the whole resampling process as best I can.
Copy link to clipboard
Copied
As long as you're working with uncompressed audio, I wouldn't worry about changing sample rates from 48kHz to 44.1kHz or vice versa too much. I'd worry a little bit more about bit depth changes, so 24-bit to 16-bit or (hopefully not) 8-bit. When it comes to changing sample rates, pitch changes are usually the thing to keep an ear out for (PR, AE and AU shouldn't introduce any).
If you have control over it in your workflow, it wouldn't hurt to maintain sample rate and bit depth at the highest settings possible from source footage to edit settings to edited master. So, if receiving a music mix for a music video at 44.1kHz, you could set your Sequence settings to match and export to that as well. But if you left the Sequence at 48kHz, the slight up-sample won't hurt anything (it just won't add anything) as PR cleanly handles the sample rate change.
As far as YouTube and Vimeo go, you could worry a little bit about introducing AAC compression at 320 kbps; however, as long as the source is uncompressed 48kHz or 44.1kHz or 32kHz 24-bit or 16-bit the resulting file should sound pretty good.
-Warren
Copy link to clipboard
Copied
Thank you.
Just a note though - in advanced audio APPs like RX - you will see that resampling can create unpleasant peaks. And so I would always use a standalone app to convert to 48k, if that's what the target destination required - and then bring that 48k file into Premiere. I'm not sure I trust Premiere to handle audio resampling in this most elegant way. I may be wrong but I see no options with regard resample... for example, filter steepness or post limiter options to prevent clipping!
The good news here is that - I can stay at 44.1k - from the source file through to upload onto Youtube et al.
I hope this stuff becomes wider knowledge
Copy link to clipboard
Copied
Is this a widely known fact?
I didn't know.
Find more inspiration, events, and resources on the new Adobe Community
Explore Now