Recorded at 44.1 -- master at 48 kHz for Soundcloud?

[Billy86]

Soundcloud says: "We recommend you to upload in a lossless format like WAV, FLAC, AIFF, or ALAC. If you can, the bit depth and sample rate should be 16 bit and 48 kHz respectively." 

QUESTION: The song is recorded at 44.1 kHz/24bit. I'm mastering it in Ozone 8. I can set the sample rate and bit depth of the output. Does it make sense to save the mastered .wav file at a sample rate of 48 kHz? I understand the reasoning behind 16 bit for streaming. But if I recorded everything at 44.1, can I/should I upsample to 48 kHz for the render? At 44.1, the audio resolution is less than what recording at 48 would have produced. So is there anything for upsampling to upsample? Is there any "there" there?

Thanks in advance. 

 

[Craig Anderton]

I wouldn't worry about it. If SoundCloud wants 48 kHz, then render the output at 48 kHz. The quality of Ozone's sample rate conversion is such that no one will hear a difference anyway.

[CJ Jacobson]

1 hour ago, Billy86 said:

So is there anything for upsampling to upsample? Is there any "there" there?

No, because there is nothing there original. You cannot add something that doesn't exist in the original format. so going form 44.1 to 48kHz will add nothing to your audio file and it will not take away nothing, except for this:  It will be a bigger file and take up more disk space. 

[John]

I see no reason to change the sample rate. I have no idea why they think its a good idea. 

[chuckebaby]

ya if it was 48 to begin with and you uploaded at 48 I could see why, but your gaining nothing by uploading a 44.1 to a 48.

[razor7music]

It would be an interesting experiment to see if "resampling" a 44.1k audio file to 48k changes the file size. Hmm...

[scook]

It does not matter how wav files are created. The disk space used is a function of the record length, sample rate, bit depth and number of channels.

44.1 kHz 16bit Stereo uses 10.1 MB of disk space per minute

48 kHz 16bit Stereo uses 11.0 MB of disk space per minute

Source: http://www.cakewalk.com/Documentation?product=Cakewalk&language=3&help=AudioPerformance.16.html

 

[razor7music]

36 minutes ago, scook said:

It does not matter how wav files are created. The disk space used is a function of the record length, sample rate, bit depth and number of channels.

44.1 kHz 16bit Stereo uses 10.1 MB of disk space per minute

48 kHz 16bit Stereo uses 11.0 MB of disk space per minute

Source: http://www.cakewalk.com/Documentation?product=Cakewalk&language=3&help=AudioPerformance.16.html

 

Yes, thanks. My interest is if by resampling a wav file that was created at 44.1k to 48k are you increasing the file size? I know an original file, with all other parameters being equal, recrded at 48k will be larger than one recorded at 44.1k. It's the conversion that would be interesting to know. For me, increasing a file size with no other purpose than Soundcloud said so, would be a negative--especially when uploading (if indeed that happens).

[scook]

13 minutes ago, razor7music said:

My interest is if by resampling a wav file that was created at 44.1k to 48k are you increasing the file size?

yes by 48/44.1

The space requirement formula does not change based on how the file is created.

[Byron Dickens]

Why on earth would Soundcloud do something like that? 48kHz is the standard for video. The standard for CD is 44.1 .

[Craig Anderton]

6 hours ago, Byron Dickens said:

Why on earth would Soundcloud do something like that? 48kHz is the standard for video. The standard for CD is 44.1 .

Good question. I suspect it's because some people experience "the Spinal Tap" effect, where if an amp knob goes up to 11, it MUST be better than an amp knob that only goes up to 10. I've met quite a few people who consider it a more "professional" rate, so maybe it's just a psychological thing. Regardless, I'm a "the customer is always right" kinda guy when you're handing off files. If someone wants 48 kHz, I give them 48 kHz. 

[razor7music]

11 hours ago, Craig Anderton said:

Good question. I suspect it's because some people experience "the Spinal Tap" effect, where if an amp knob goes up to 11, it MUST be better than an amp knob that only goes up to 10. I've met quite a few people who consider it a more "professional" rate, so maybe it's just a psychological thing. Regardless, I'm a "the customer is always right" kinda guy when you're handing off files. If someone wants 48 kHz, I give them 48 kHz. 

What if they want vinyl? ? jk

Edited May 3, 2019 by razor7music

[garybrun]

44.1 was all because of CD.

it would be good to have now a standard of 48 that’s also matches video projects?

[Craig Anderton]

7 hours ago, razor7music said:

What if they want vinyl? ? jk

Well, then I give them a file mastered for vinyl...having been brought up in vinyl world, I know how to do it

 

[Shayne White]

SoundCloud used to allow 96KHz uploads, which is what I prefer. Is that not possible anymore? 44.1KHz is an obsolete format in my opinion.

[Craig Anderton]

1 hour ago, Shayne White said:

SoundCloud used to allow 96KHz uploads, which is what I prefer. Is that not possible anymore? 44.1KHz is an obsolete format in my opinion.

96 kHz can make a major improvement for sounds generated inside a computer, like virtual instruments and amp sims, as well as limiters with super-fast attack times. This is because it eliminates foldover distortion (which BTW 48 kHz will not do as effectively). So it's easy to make a case for recording and mixing at 96 kHz, because of the improvement in sound quality. Sonar was the first DAW to recognize this by including upsampling, which I had suggested to Noel and he not only ran with it, but simplified how to do it.

However...in terms of a delivery medium, I've done a lot of classical sessions at 96 kHz with delicate acoustic sounds, reverb tails, etc., as well as some rock and EDM. I have yet to find anyone who can reliably (better than chance) differentiate between audio recorded at 96 kHz and played back at 96 kHz, vs. audio recorded at 96 kHz and played back at 44.1 kHz. So when it comes to a playback, delivery medium, I think Nyquist nailed it. But when it comes to recording, 96 kHz is a good way to prevent foldover distortion.

I'll take it one step further. 192 kHz can sound even better than 96 kHz for sounds generated "in the box" (again, if something acoustic or electric is coming in via an audio interface, it doesn't matter). But doing a multitrack session at 192 kHz is highly impractical. It makes a lot more sense to do it at 44.1 kHz (or 48 kHz or 96 kHz, if that's how you roll), and selectively upsample those parts that benefit from higher sample rates. For more on the topic, check out the article Software Tech: 96 kHz vs. 44.1 kHz—Let’s Settle This. 

For a demo of the difference that higher sample rates make to virtual instruments, check out this video.

 

[Billy86]

Thanks for weighing in everyone. I can’t figure out any reason how/why a tune recorded at 44.1 would benefit from being rendered at 48.  I’ll go with 44.1  

Razor7, I’ll also output at 48 just to see if the file size differs and post what I find.

 

[bitflipper]

1 hour ago, Billy86 said:

I can’t figure out any reason how/why a tune recorded at 44.1 would benefit from being rendered at 48.

That's because there is none. Think about it: you use a higher sample rate because it contains more information. After upsampling an existing file, where exactly would that additional information come from? It can't be manufactured out of thin air.

[Craig Anderton]

The ONLY potential improvement I can think of, and it would be such a slight improvement as to be insignificant, is that when playing back a 48 kHz file, the playback system might be able to use a gentler smoothing filter that might make for a slightly better sound. But that's a real stretch, because the difference between 44.1 and 48 kHz is so small. And as Bitflpper points out, the file itself won't be any better. 

[Billy86]

Just to answer Razor's question:

Rendered at 44.1: File size is 73.7 megs

Rendered at 48: File size is 80.2 megs

I can hear no difference.