Skip to main content

As requested, here is a video testing SoundCloud's conversion quality.
Let me know what you think

Comments

audiokid Tue, 08/14/2018 - 09:13

Nicely done as always.

In my original tests, I noticed files too loud are more prone to sounding wrong. Too loud mixes always sound less focused, and will not render well as mp3 .
Thank you for mentioning that.

It would be interesting to do this test with a mono file. (See how a mono mix converts and renders).
I think people need to check their mixes for phasing issues prior to mastering or mixing down.

kmetal Wed, 08/15/2018 - 18:04

nice work man! looks like you've brought the truth once again, just like when you did the DAW sonic comparison. im shocked with this one that there was little or no difference at all that i could hear between any of them. so i guess its pretty safe to say SC is ok, its just poor mixing doing it. the only two remaining questions i have would be what happens when you upload a bad mix, (since yours was good), and if you used a third party, like sonnox fraunhauffer conversion to the SC bitrate, if there would be any noticeable difference between the files. although im not sure its possible to do to that, ie if SC out there stamp on everything.

pcrecord Thu, 08/16/2018 - 05:31

kmetal, post: 458575, member: 37533 wrote: the only two remaining questions i have would be what happens when you upload a bad mix, (since yours was good), and if you used a third party,

That's the thing K, I tried offsetting the overheads and the Hihat track to introduce phase issues and I also tried to make a mix way too loud.. but it seems to be still ok.
I'm suspecting Pro L2 is doing a too good job but I'm not sure what I can do next. I tried using cakewalk too.
To this day I didn't find the point of faillure to Soundcloud. If the problem comes from the mix, I guess mixing with good technic like I do ain't going to cut it !
I'm open to suggestions and ready to make a part 2 !

So far, I've been ask for null tests, to test with mono files and I could also go to mp3 128kbs,
It could also be that my conversion software does a better job than others too.. lol

kmetal Mon, 08/20/2018 - 16:59

pcrecord, post: 458580, member: 46460 wrote: That's the thing K, I tried offsetting the overheads and the Hihat track to introduce phase issues and I also tried to make a mix way too loud.. but it seems to be still ok.
I'm suspecting Pro L2 is doing a too good job but I'm not sure what I can do next. I tried using cakewalk too.
To this day I didn't find the point of faillure to Soundcloud. If the problem comes from the mix, I guess mixing with good technic like I do ain't going to cut it !
I'm open to suggestions and ready to make a part 2 !

So far, I've been ask for null tests, to test with mono files and I could also go to mp3 128kbs,
It could also be that my conversion software does a better job than others too.. lol

your just to damn good at what you do marco. the only thing i could think of would be to try a waves L2, or what im wondering is, if we were hearing artifacts related to 32 bit systems? i also wonder if the tracks were recorded using something like a scarlett, a low end conversion, if that would make a diff.

i honestly am starting to believe this video is proving to me that SC is perfectly adequate for what it is, and its operator error, thats ultimately causing the issues. your vids are highly informative! still love the one where you proved DAWs sound different.

pcrecord Tue, 08/21/2018 - 08:55

kmetal, post: 458615, member: 37533 wrote: your just to damn good at what you do marco. the only thing i could think of would be to try a waves L2, or what im wondering is, if we were hearing artifacts related to 32 bit systems? i also wonder if the tracks were recorded using something like a scarlett, a low end conversion, if that would make a diff.

Yes, I'll try to replace Fabfilter Pro L2 by waves L2 and see if it makes a difference.
I could also use the stand alone version of Ozone without other external plugins.. Maybe Samplitude is too well built to let out screwed files.. ;)

kmetal, post: 458615, member: 37533 wrote: i honestly am starting to believe this video is proving to me that SC is perfectly adequate for what it is, and its operator error, thats ultimately causing the issues. your vids are highly informative! still love the one where you proved DAWs sound different.

I'm guilty for saying Soundcould screwed the audio on a few posts here on RO. I always assumed because when posted directly in mp3 I heard some problems go away.
But now I'm thinking they might have changed their conversion algorithm or like many thinks the conversion just worsen problems already present..

KurtFoster Tue, 08/21/2018 - 13:14

it's placebo effect. it's easy to say now but i have suspected it all along.
but then the whole idea of the internet is to get files small enough to be transfered quickly. speed vs. quality. imo the average cassette of yesterday sounded better than any mp3. i've heard an mp3 of the Doors song "Soul Kitchen" where the guitar sounded like a banjo. and it wasn't a banjo.

imo it's moot because it's mistake to make any judgment based on any mp3 other than "this is what your song sounds like when it's rendered into an mp3".

still there is merit to requesting that files be loaded here instead of a link to any other site as far as long term archives.

pcrecord Tue, 08/21/2018 - 16:15

Kurt Foster, post: 458635, member: 7836 wrote: it's mistake to make any judgment based on any mp3 other than "this is what your song sounds like when it's rendered into an mp3.

This is a wise man talking right there !!

I manage to upload 2 more files using samplitude and Waves L2.. One uploaded to Soundcloud at 16bit44khz and the other a mp3 of 128kb/s (I know it's streching but I wanted to hear the difference)

[MEDIA=soundcloud]pcrecord/theglass-waves-l2-1644

[MEDIA=soundcloud]pcrecord/theglass-waves-l2-mp3-120kbs

kmetal Tue, 08/21/2018 - 16:45

pcrecord, post: 458630, member: 46460 wrote: But now I'm thinking they might have changed their conversion algorithm

this is quite possible

pcrecord, post: 458630, member: 46460 wrote: I could also use the stand alone version of Ozone without other external plugins..

id suspect that would end up artifact free as well. it is pssible i belive to use 3rd party plugs in ozone standalone, but i belive waves pluggins are some of the few incompatible plugs with ozone standalone.

Kurt Foster, post: 458635, member: 7836 wrote: still there is merit to requesting that files be loaded here instead of a link to any other site as far as long term archives.

i agree. i wish we could use .wav here, or maybe a better quality compression, i think OGG is one of them? bandcamp rules the roost for allowing people to upload about half a dozen different formats, wav and aiff being two of them. im not sure if there are sample rate restrictions. obviously band camp if a different thing than RO, but just mentioning it for conversation.

considering most artists are tracking and mixing at 96k, and have been for a decade, i thinks its time to get rid of 44.1 as the idea of 'standard'. theres definitely dithering and data loss, moving from 96k to 44, audible or not. with HDD space being massive, and cheap, and SSD, and NVME SSD being pretty affordable, it really is time to move away from data compressed audio. Dropbox streams .wav files with hardly any load time, to any old phone or device. and thats on a 10gbps wifi connection. around my area, sCamcast interent provider is offering gigbit for $89 per month, so between gigabit internet, and 1500mbps+ storage, we should start to hear full quality audio

as an asidei own the domain name PerformerCloud.com, and i hope to offer a free or cheap cloud service for HIFI audio. its time. i hope to have my sites up and running by 2020 or before.

-- edit-- sorry i wrote this reply before marco posted his new samples, cant wait to have a listen.

pcrecord Thu, 08/23/2018 - 06:22

kmetal, post: 458647, member: 37533 wrote: yeah man i think you are on to something. im not sure if its the bit rate, the L2 or both, but they really sound full of artifacts. i can imagine that many many SC users are uploading L2'd 128kbps files. i think your starting to shed the light on this! looking forward to part duex.

I'm on it. As soon as I get time, I'll do a few more tests as asked, mono null tests, with Fab pro L2 and I'll export them all to 16bit44khz and then converted to mp3 320kb/s then 128kb/s
It will make a good part2, I'm sure !

audiokid Fri, 08/24/2018 - 20:56

pcrecord, post: 458642, member: 46460 wrote: This is a wise man talking right there !!

I manage to upload 2 more files using samplitude and Waves L2.. One uploaded to Soundcloud at 16bit44khz and the other a mp3 of 128kb/s (I know it's streching but I wanted to hear the difference)

[MEDIA=soundcloud]pcrecord/theglass-waves-l2-1644

[MEDIA=soundcloud]pcrecord/theglass-waves-l2-mp3-120kbs

Looking forward to your findings and comments. It would be interesting if you included research to null them all and study the remainders including what happens to the left and right or mono.

Also, to my ears there is always a sonic quality difference with how an mp3 processes one file to another exact mixdown with even .01 db of capture change prior to mp3 (master) compression.
But to do this you need a second DAW to capture the mixdown. I've noticed mp3 always sound better when I find the sweetspot capture level and then use a limiter to increase level just below -01. I wonder why.

audiokid Fri, 08/24/2018 - 21:04

Makzimia, post: 458676, member: 48344 wrote: When I was using SoundCloud personally, all my songs went up as wav files, never MP3.

And my MP3 songs up on here have all been 320k

I've compared my Mp3 to SC and as long as I have a quality mixdown, everything in phase and transients controlled but not smashed, and use a limiter at the end of the master, I don't hear enough of a difference choosing one way or the other. To me its all about mixdown integrity, level prior to the master.

cyrano Sun, 09/16/2018 - 14:17

I'm wondering if the people who find Soundcloud and other streaming services' quality not so good, take their connection speed into account.

Streaming servers will deliver different bitrates for different speeds, so people on a lesser connection will be able to hear the music.

I'm not sure if this affects audio a lot, but I'm pretty sure it affects video. You can demonstrate that quite easily, by starting several Youtube video's on the same connection, preferably from different devices.

KurtFoster Sun, 09/16/2018 - 15:03

for most the primary function of Sound Cloud is to get their music exposure. if you are concerned with audio quality, i submit that mp3s and posting on the internet is not the place or the method. 99.999 percent of the users on Sound Cloud are listening on a phone through ear buds. we're the only ones who are nuts enough expect quality reproduction. i miss my Tascam 40-4 and 2A mixer. it sounded better to me than any digital pos.