I had considered building a SoundCloud clone a few years back, but dismissed it for pretty much the same reasons as mentioned in the blog. The cost of the encoding, storage and CDN endpoints was just too much for even a rudimentary song sharing service.
NB: I use SoundCloud to share my own recordings and song ideas, and to listen to other people's ideas and do collaborations with them. NOT to listen to mainstream music or podcasts. I wanted my clone service to facilitate that.
Which is why I had never really factored in copyright violations. But I am thinking as well as human moderation, couldn't such a service also employe a HackerNews style voting system - basically if a song started getting flagged by users as 'copyright violation', it would start to disappear down any searches or catalogues, until enough votes caused it to disappear from the service altogether?
>couldn't such a service also employe a HackerNews style voting system - basically if a song started getting flagged by users as 'copyright violation', it would start to disappear down any searches or catalogues, until enough votes caused it to disappear from the service altogether?
depends what your users' interests are. if they genuinely want the site to only offer legal music, then sure it might work. but if most of your users are using the site as a free streaming service, it's not in their interest to remove music that they like.
Even youtube is suffering from blatant copyright violations with all those piracy live streams. I still don't completely understand why that suddenly became a thing.
If you gave up because of the cost, your reason is different from the blog. Or I did not understand the post.
About copyright, why not make it difficult to find protected music? If people share their own content, the name of this file will not be named as a Justin Bieber song. It will be your own name. I believe most sites want to share protected music and then try to negotiate their use.
Eventually, people may try to game the system, but the site may have an automatic way to freeze the song after receiving a complaint and check if it is a copyright infringement. And this file can be used to feed the system, so it can avoid futures complaints.
Well, perhaps I should clarify that I gave up because my assessment of the costs paralleled the research and numbers in the blog post. No way could I run this as a 'side project' on beer money, because if the site became even remotely popular, the escalating costs of the hosting, CDN, encoding etc. would have just slammed me quickly, and I didn't have any monetisation strategy in place.
I basically wanted to build what SoundCloud was in its early days, before they lost touch with their user base.
The other main point about the copyright infringement in the article wasn't on my radar at all, because I wanted my site to be for music creators to share ideas and collaborate, rather than post commercial music on - but thinking back, THAT could have become a huge problem too, if, as you have suggested, nefarious people had simply used my platform to share non public domain music.
I think it's interesting that pretty much every major media company has had to invent their own ContentID and that there are very few companies in the copyright space. Actually, the two companies I knew of were both run by the same founders and both acquired pretty quickly (Rightsflow acquired by Google/Youtube and Source3 acquired this past week by Facebook)
The fundamental reason why none of the streaming services will remain in business is market economics. Digital Music is a commodity because Apple set an unrealistic consumer price point against which no independent service can compete.
Consider that Apple sets the market price for music, from a consumer perspective, and at a flat $0.99 per track with some tracks priced higher and lower. For brevity, I'm going to use the $0.99 price. This price does not reflect actual costs for providing the service, rather, it is a price point to encourage adoption by the masses. At $0.99 per track one can argue that Apple was seeing maybe 5 to 7% net margin.
Any competing company could not set the price higher than $0.99 as the audience would just go to Apple for the lower amount. So as a competitor; e.g., PressPlay who became the legal Napster, would need to compete at an 7% net margin which is extremely low. Businesses generally can't survive on less than a 10% net margin and most aim for 20% to 40% net margin.
How does Apple get away with running a business at essentially a loss? They sell ancillary products (hardware) at very high margins and the music is the loss leader that makes those products more valuable.
Apple made music a commodity and Google Music and Amazon Prime Music just reenforce the situation by essentially charging you near nothing for unlimited music.
Streaming services, by comparison, have very little they can sell in addition to their subscription fees, and those fees cannot be much higher than what Apple, Google, Amazon would charge for similar services. The phrase "they'll make it up in volume" holds no water when licensing fees scale in proportion to usage. Just look at P's 10Q for reference.
BTW, I was involved in building or supporting the following music services in one way or the other back the early 2000's:
* MP3.com
* Trusonic
* Pressplay -> Napster under Roxio
* eMusic
* Rollingstone.com
* MP4.com
Since 1999 it was clear that streaming was the future of music. Just not a very profitable future as it turns out.
If you responded to the copyright takedown request(s) in an automated fashion, you might have a chance.
I don't know if it's a requirement for you to police it. Only to respond to the DMCA requests. So make the website, under a corporation (which is easy to do), then build a system which auto takes down pending moderation for every DMCA request.
Obviously, this won't scale immediately, but use mechanical Turk to review the requests initially. Once you know content is copyrighted, keep it. Use it to compare against other uploads.
YouTube doesn't screen every copyrightable material.
Once you get started, get some money, then try to make deals.
Let's see. This guy's thesis seems to partially be that "SoundCloud's paid user tier is a ripoff because the raw cost of the expanded storage you get is actually negligible".
What else might a paid user's funds be going to, that he's not thinking of?
* paying for the time of people who maintain and develop the site
* paying for the time of people who support the people who maintain and develop the site (receptionist, office manager, etc)
* paying for the time of people to deal with copyright issues - both in terms of having people whose job is to build and maintain the automated copyright infringement detection systems (and to liase with the recording industry people they need to), and in terms of maybe having some kind of batch of people whose sole job is to do it manually
* possibly paying for the time of other content monitoring humans, doing the equivalent of the poor bastards whose job it is to monitor facebook/youtube/flickr/whatever for child porn, snuff, and other horrible things, is there an audio equivalent to that?
* office rent (SoundCloud apparently has offices in Berlin, New York, London, San Francisco, Los Angeles & Sydney, none of these are exactly cheap places to rent - and that drives up the salaries of your workers as well)
* bandwidth to play the songs for the people who are getting a lot of views
* advertising SoundCloud
* perhaps some measure of profit-sharing with the users who generate popular content (I don't think SoundCloud has ads, though? Looks like it does, I guess I've never spent enough time using it as my music source to hear any.)
* maybe paying back the VC/founder money that's been put into getting the site to where it is
* who knows what kind of financial arrangements with the music industry, I can't even begin to speculate how complex those get
Sure, the carrot SoundCloud holds out in front of you of "more storage space" is negligible in value. But don't neglect all these other costs. Money for those doesn't just come out of thin air. Unless they've figured out a way to massively parallelize mining BTC with the spare CPU cycles of everyone listening to a song or something.
(Limiting the upload bandwidth of free users may also be a deliberate choice to keep the quality of SoundCloud's content slightly better on average. If you're not making enough money off your music to feel like $7/mo is worth it then it's probably not very good yet to be honest. But I'm getting off into total theorizing here and I'm not even stoned.)
I think SoundCloud as a service to the consumer is great. Not so much on the business side... Which sucks because so much of the music in my playlist today I found originally through SoundCloud.
They also update to use SC on the backend. They used to host the songs, but now it's some agreement thing with SC.
Was a die-hard HypeM user, and they lost me when they partnered with soundcloud bc offline listening was the _best_ feature of the app in an NYC subway...
I'm not sure if it helps, but a while ago I created PodCloud to generate RSS feeds for SoundCloud users, allowing me to use any podcast client to automatically download new releases and listen to them offline. Haven't touched it in years, but it seems to be still working.
> You can't detect copyright infringement without access to the original copy.
This sounds like an interesting problem. Somehow, buy borrow or steal the top 10,000 (?) tracks on spotify so you can finger-print them.. hmm probably more like 100,000 tracks.
After reading what Warner did to SoundCloud: it's nauseating to imagine that I end up feeding and nurturing monsters like that company and other music labels when I subscribe to Apple Music and enjoy my favourite songs :-(
NB: I use SoundCloud to share my own recordings and song ideas, and to listen to other people's ideas and do collaborations with them. NOT to listen to mainstream music or podcasts. I wanted my clone service to facilitate that.
Which is why I had never really factored in copyright violations. But I am thinking as well as human moderation, couldn't such a service also employe a HackerNews style voting system - basically if a song started getting flagged by users as 'copyright violation', it would start to disappear down any searches or catalogues, until enough votes caused it to disappear from the service altogether?