What Kind of File Should I Upload to Soundcloud

While some streaming services similar Amazon Music Hard disk drive and Tidal are now offering lossless audio, many others similar Spotify, Apple tree Music, and SoundCloud still use lossy audio compression techniques to deliver music. Of those, SoundCloud has always been unique in how easy it makes instant uploads for creators.

Mayhap it'southward due to that very ease that questions like, "Why does my music sound unlike on SoundCloud?" or "What can I exercise to brand my music sound better on SoundCloud?" seem to come up more often than they do for other streaming services.

Despite SoundCloud introducing a new "mastering" feature to optimize streaming playback, knowing what really happens to your audio during streaming and mastering is key to understanding how to produce a track with the highest possible sound quality for streaming. So let's take a look at why those sonic changes occur, and what we can practise to minimize them.

In this piece you'll learn:

• How to optimize your songs for streaming on SoundCloud and other compressed audio formats

• What you can and can't control in the process

The bottom line

To become to the bottom of this, I prepared 40 masters of a single song—xx at 44.1 kHz and 20 at 48 kHz—and uploaded them all to SoundCloud. For each sample rate, I methodically varied the parameters of peak level, crest factor, frequency-specific width, and total width. I then played them all back off SoundCloud, recording the output bitstream pre-conversion—again at 44.i kHz and 48 kHz—for assay and comparison confronting the originals. This yielded a whopping eighty versions of the song!

• twenty uploaded and recorded at 44.1 kHz

• 20 uploaded at 48 kHz and recorded at 44.1 kHz

• 20 uploaded at 44.one kHz and recorded at 48 kHz

• xx uploaded and recorded at 48 kHz

Testing xl versions of a song

After level matching them all for a fair comparing, I got to work listening and measuring to determine which factors played the biggest role in preserving—or degrading—sound quality during format conversion and streaming playback. At the cease of the day the parameter which made the biggest impact was: width! Not only that, simply all the other variables had picayune to no impact (caveats ahead).

To understand why this is, how you can potentially have reward of it, and why you might non want to worry most it at all, read on!

Manipulating width for a "better" encode

I should authorize what I mean by "better." Really, what we're talking about is an encode which is perceptually closer to the source. Nevertheless, the steps nosotros're taking to get there involve making some sacrifices to the source. So while the encode and the source may sound more alike, the cumulative divergence between the encode, the source, and what you were originally trying to attain may still be fairly noticeable.

That qualifier aside, hither are a few things y'all can practise to minimize the differences between the source and the encode:

Narrow the high-end

Using a tool like the Imager in Ozone 9, try narrowing frequencies in a higher place about 8 kHz. I can't give yous a precise amount, as it will very much depend on the corporeality of width that you had in that range to begin with. Endeavor soloing that ring and reducing the width until it occupies about half of the stereo field between your speakers. This will assistance reduce some of the loftier-frequency washiness that is so common with low-bitrate lossy codecs.

Narrow mid and low frequencies

If you want, and your primary tin can handle it, try narrowing the mid and depression bands every bit well. Try setting the mid ring to well-nigh 1–8 kHz, and the depression band below 1 kHz. You lot could even split this into two ranges: 400–1000 Hz and below 400 Hz. You'll probable want to leave the mid—and low-mid if you're using it—bands fairly shut to their original width, however, you may be able to get away with narrowing lower frequencies a flake more. Any little bit helps.

Use a mono primary

This is absolutely an extreme solution, but if y'all can justify it, a mono source will give you the "all-time" encode—once more, meaning perceptually closest to the source, albeit now in mono. This is because you're essentially asking the encoder to do half as much work by encoding a single channel. In turn, this means the encoder can allocate it's unabridged bandwidth to that one channel, rather than having to divide it betwixt two channels.

The reasons width plays such a disquisitional role in encoder performance are hugely complex, but can be summarized as follows: almost lossy encoders similar AAC, MP3, and Opus utilize a technique known as joint stereo encoding. This ways that rather than encoding both left and right channels independently, they employ multiple techniques such as mid/side and intensity-stereo coding to optimize bandwidth allocation to where it will be well-nigh noticeable—ofttimes the center of the stereo paradigm.

The end result is that ultra-broad stereo signals often suffer from quality deposition more than noticeably than do narrower ones. Additionally, high frequencies require more bandwidth to encode. Thus, by reducing the width of high frequencies, not merely do you free up some bandwidth for the encoder, allowing it to classify its bits more efficiently, but you also foreclose some of the more noticeable, warbly, washy distortion from showing up in the encode.

A great fashion to experiment with the effects of these changes in existent-time is by using the Codec Preview in Ozone 9 Advanced. Endeavor using MP3 at 128 kbps or AAC at 256 kbps—ii of the common codecs used by SoundCloud depending on the playback platform and subscription level—and tweaking Imager parameters. You can even use the "Solo Artifacts" role to hear how changes in width impact the underlying distortion added by the codec.

Codec Preview in Ozone 9

All the other bits

I would exist remiss if I didn't address things similar peak level, crest-cistron, and file format for upload, so let's talk about those at least a little.

In all my recent tests, top level did not have a noticeable touch on on encoder operation—at least not directly. Past this, I mean that then long equally there wasn't any clipping, the encoder functioning between versions with dissimilar amounts of top headroom was identical.

However, because lower bitrates—such as those often used by SoundCloud—can cause peak level overshoot of a decibel or more, it's skilful practice to set the ceiling of your limiter to -ane or -1.5 dB and utilise a True Peak limiter such as the Ozone Maximizer. This helps forestall clipping on playback, especially through cheaper consumer devices.

The story with crest gene is largely the same. While it doesn't have a direct, dramatic impact on encoder performance, a lower crest factor will often effect in higher pinnacle level overshoot—something which ultimately oftentimes results in DAC clipping and distortion. This has the slightly ironic event of requiring additional pinnacle headroom—or a lower limiter ceiling—the college you button your boilerplate level, something which can speedily turn into a losing battle.

This is another area where Codec Preview in Ozone nine Advanced tin exist enormously helpful. By turning on Notice "True Peaks" in the I/O options and listening through the MP3 128 kbps codec, you can fine-tune the Maximizer threshold and ceiling to attain an optimal level while avoiding post encode clipping.

Checking post-encode peak headroom in Ozone ix

Every bit for upload format, the official recommendation from SoundCloud is a 16-bit, 48 kHz WAV file. This reason for this is that of the several codecs used, the majority of them are set to take in a 48 kHz file, then this minimizes the amount of sample rate conversion that volition take place.

That said, sample rate conversion has go extremely transparent, and in my tests neither the upload nor playback sample rates had an appreciable effect on encoder performance or playback quality.

The 1 caveat here is that if yous enable downloads on SoundCloud, the file you upload is the 1 your fans go when they download. Thus, if you want them to receive a 320kbps MP3, that's what you lot'll need to upload. However, this results in transcoding from ane lossy format to some other, which never sounds particularly adept.

In short, if you want the best streaming quality possible, upload a 16-bit WAV at 44.i or 48 kHz. If, on the other hand, yous want to enable downloads, upload the file yous want your fans to receive, only know that if it's a lossy file, streaming quality volition suffer. Since these days downloading a local copy is probably not as common as information technology once was, this may be a moot point.

Decision

To wrap up I desire to consider a few reasons why possibly you lot shouldn't worry too much nearly all the factors nosotros've just discussed.

Commencement and foremost, SoundCloud may well update the codecs they use in the future just as they have in the past. When that happens they will re-encode all uploaded music to take advantage of the new codec(s). It's for this very reason that they themselves urge creators non to endeavour to optimize files likewise much for a specific codec.

2nd, while you tin can control the width, sample rate, etc. of the file you lot upload, you tin can't control how your fans will listen to information technology. Of class, this is true of the vast majority of playback mediums. Information technology bears repeating here though because fifty-fifty on SoundCloud alone, the playback experience can vary depending on subscription level and playback device. Consider carefully whether it's worth sacrificing some of the width and spaciousness of your track just for the everyman mutual denominator.

Hopefully, this has armed you lot non but with some of the tools to ameliorate encoder functioning when uploading to SoundCloud but also the wisdom to know when, when not, and how strongly to wield them. Good luck, and happy mastering!

burgerfrount.blogspot.com

Source: https://www.izotope.com/en/learn/mastering-for-compressed-audio-formats.html

Share this post