For distributing music online: 44.1 kHz or 48 kHz?

I know the whole sample rate debate is severely overblown. It's quite possible that, in all practicality, this situation is a nonissue.

I just want to clear up a few concerns so that I can make a confident and educated decision and lay uncertainty to rest.

Spotify always converts to WAV 44.1 kHz, so I'm highly inclined to record and bounce at that sample rate...

However, my concern with 44.1 kHz is the issue of anti-aliasing, that 44.1 kHz might be sonically subpar (whether that means distortion on the song when played back at lower quality formats or maybe loss of very high-end frequency content in the song, I'm not entirely sure) to 48. I also read a post David made talking about how the pros work at 48 kHz, and the hits that play on the radio are at 48. I don't want to record and bounce at 44.1 kHz only to find 48 would've put me on par fidelity-wise with professional releases (assuming the much more important aspects of music making--production, recording, mixing, mastering--were on par as well).

Based on what David said and other good things I'm reading about 48 kHz (such as it's useful if you ever have your song on TV/in a movie), I figure I'll record/bounce at 48 to be safe...

However, I'm reading that SRC and downsampling is often less than ideal, and can result in loss of audio fidelity, especially without expensive equipment and converters.

Is this SRC and downsampling concern of mine a nonissue, and I should just bounce down to 44.1 (from recording/working at 48) when I export my WAV from Logic?

Or does recording at 48 kHz really not provide sufficient fidelity-related improvement to justify itself, and I should just record at 44.1 to avoid the potential problems of SRC?

Thanks to anyone for any input.

I'm not sure what Spotify are doing or how they are doing it, but for me, I'd probably want to at least take charge of the process, rather than expect some third-party company would treat my material with the appropriate respect.

So, if you are dead keen to record at 48, then do so. If you're then concerned that some services SRC to 44, then *do the SRC yourself* with an appropriately good convertor you can audition and be happy with, and provide those services with the 44 version.

This also gives you the additional advantage that you can A:B the 48 and 44 versions, and make sure you're happy.

This also means that you can *test this yourself*, to verify how destructive a process the SRC is on your material. In the early days of digital back in the 90s, SRC's weren't that great. Nowadays, there are really good ones, and you should try a variety to satisfy yourself of how it sounds.

If, in your tests after all that, you *aren't* happy with SRC from 48 to 44 - then you now know you want to use 44 in your projects, set it and forget it, and you won't have to do any SRC at all.

As always - you really need to do these things yourself, rather than have people tell you what you should do. The above information should give you a useful guide of things to audition while you develop your own workflows for your material. I have my own feelings on the issue (as we've discussed already ) but you need to find what works for you, really.

And do let us know what you find works for you best, in the end, for other folk with the same questions...

I'm not sure what Spotify are doing or how they are doing it, but for me, I'd probably want to at least take charge of the process, rather than expect some third-party company would treat my material with the appropriate respect.

 

So, if you are dead keen to record at 48, then do so. If you're then concerned that some services SRC to 44, then *do the SRC yourself* with an appropriately good convertor you can audition and be happy with, and provide those services with the 44 version.

 

This also gives you the additional advantage that you can A:B the 48 and 44 versions, and make sure you're happy.

 

This also means that you can *test this yourself*, to verify how destructive a process the SRC is on your material. In the early days of digital back in the 90s, SRC's weren't that great. Nowadays, there are really good ones, and you should try a variety to satisfy yourself of how it sounds.

I'll definitely do some A/Bing. Say I join the 48 kHz club and end up having to find an appropriately good converter to audition since the downsampled export from Logic sounded bad, do you have any recommendations where I might start? Which dedicated hardware or software converters do you use, if any, and why?

I would record at 48kHz and convert to a 44.1kHz wav file for spotify upload.

Thanks for the straightforward answer. Have you had an experience or heard of one where audio problems were created from exporting a 44.1 kHz WAV file from a Logic project which was all recorded and worked on at 48 kHz?

Personally i'd output in the default that your projects are setup, or export at 44khz direct from Logic. Because streaming platforms will transcode to a compressed version no matter what format you start with anyway, and it's compressed formats where quality difference will occur. The main gains on audio compression is attained by cutting frequencies beyond the human ear anyway.

Automated services, such as spotify's upload process, will carry out the same actions regardless, i.e. a 44khz file will still pass through their standard conversion process that has been optimised over the years. Everything just runs through on a script and comes out in the format required their end. Even within the 44khz standard there's variances that exist.

If you're converting yourself it's just adding another conversion for the sake of it (IMO).

i.e. going Logic 48khz > 44Khz > Streaming Platform Conversion

vs

Logic 48khz > Streaming Platform Conversion

Just seems pointless to me, i expect plenty of professional/commercial recordings get uploaded direct @ 24/96, and pass through several companies before Spotify too.

A/B it, but if you do a true blind test i'd be surprised if you notice a difference. And if you do, i'm sure Spotify would be interested.

I would record at 48kHz and convert to a 44.1kHz wav file for spotify upload.

Thanks for the straightforward answer. Have you had an experience or heard of one where audio problems were created from exporting a 44.1 kHz WAV file from a Logic project which was all recorded and worked on at 48 kHz?

No. Sample Rate Conversation has gotten much better than the early days of digital these past few years, and introduces little distortion nowadays. There's no cause for concern.

I'm not sure what Spotify are doing or how they are doing it, but for me, I'd probably want to at least take charge of the process, rather than expect some third-party company would treat my material with the appropriate respect.

 

So, if you are dead keen to record at 48, then do so. If you're then concerned that some services SRC to 44, then *do the SRC yourself* with an appropriately good convertor you can audition and be happy with, and provide those services with the 44 version.

I don't disagree with you, BUT unless you're really providing the final product directly to a service like Spotify and not via a label or aggregator, you often have very little to no control over what is actually delivered to the service. Labels have their own "protocols" and pipelines with specific naming conventions and sometimes format changes (!) and some distributors want 24bit/44k, some 16bit, etc. I absolutely understand the OP's concern, but I'd do the best mix and master I can, provide the label/service/aggregator the highest fidelity final version they take*, then leave it, as the quality change in the possible SR conversion will not make or break your next hit record.

*) ...and if you work in 16bit 44k, NO point, or no benefit, in exporting 24bit 48/96k for eg., as that won't make your original mix any better sounding, is just technically higher fidelity file, via an unneeded conversion, you need your whole project to be those higher specs.

Personally I'm still currently using 24bit/44k, because I started with 16bit/44k back in the day, but I'm leaning on switching to 48k at some point, and hoping that the whole industry changes to 48k, not because I can hear the difference, but because it's dumb to have two standards (and I don't see film and TV industry changing theirs), it's just extra hassle (and conversion quality loss).

I'm not sure what Spotify are doing or how they are doing it, but for me, I'd probably want to at least take charge of the process, rather than expect some third-party company would treat my material with the appropriate respect.

 

So, if you are dead keen to record at 48, then do so. If you're then concerned that some services SRC to 44, then *do the SRC yourself* with an appropriately good convertor you can audition and be happy with, and provide those services with the 44 version.

I don't disagree with you, BUT unless you're really providing the final product directly to a service like Spotify and not via a label or aggregator, you often have very little to no control over what is actually delivered to the service. Labels have their own "protocols" and pipelines with specific naming conventions and sometimes format changes (!) and some distributors want 24bit/44k, some 16bit, etc. I absolutely understand the OP's concern, but I'd do the best mix and master I can, provide the label/service/aggregator the highest fidelity final version they take*, then leave it, as the quality change in the possible SR conversion will not make or break your next hit record.

*) ...and if you work in 16bit 44k, NO point, or no benefit, in exporting 24bit 48/96k for eg., as that won't make your original mix any better sounding, is just technically higher fidelity file, via an unneeded conversion, you need your whole project to be those higher specs.

Personally I'm still currently using 24bit/44k, because I started with 16bit/44k back in the day, but I'm leaning on switching to 48k at some point, and hoping that the whole industry changes to 48k, not because I can hear the difference, but because it's dumb to have two standards (and I don't see film and TV industry changing theirs), it's just extra hassle (and conversion quality loss).

Personally I'm still currently using 24bit/44k, because I started with 16bit/44k back in the day, but I'm leaning on switching to 48k at some point

That's pretty much where I am. I've stuck to 44 (for non-video related content) because it's fine and it's a habit from when CD was always the ultimate destination format (and SRC was generally regarded as something to avoid), but today, the landscape, and available technologies are different, so I'm thinking of updating some of those practices...

does radio still exist? (seriously, am asking...).

distributors like tunecore (who i use), and distrokid (who a collab prefers) require 44/16 wav files. so (for me), what works is recording 44/24, then, in mastering, dithering to 44/16... and off those wav files go.

24bit over 16bit sounds better (to my ears)... and it's my ears i trust. and since 44 is the final sample rate, i record at that rate, so... no downsampling.

go with what feels right... but check the final wav file. if it sounds good, it should be good.

does radio still exist? (seriously, am asking...).

Where i live, getting played on local radio is still very much the route upwards. Particularly if you're a gigging band.

does radio still exist? (seriously, am asking...).

Where i live, getting played on local radio is still very much the route upwards. Particularly if you're a gigging band.

got it!