I have been listening to computer-based digital music for nearly two decades now. This includes streaming internet radio as well as music files like MP3s. It’s amazing to recall my first efforts to play then-new MP2 and MP3 format files I downloaded, very slowly, over my 56kbps dial-up connection, around 1996 or ’97. In fact, my computer at the time (a first-generation Pentium class PC) couldn’t even handle the computational overhead needed to play back an MP3 file smoothly.
These compressed files were fascinating to me because they represented the first practical way to distribute something that resembled high fidelity audio files over the internet. Sure, Real Audio and Windows Media Audio were also early, and successful, methods for streaming audio to listeners who were mostly accessing the ’net at glacial dial-up speeds. But over a modem their sound quality was closer to shortwave or AM than FM–good for talk, but not so great for music.
MP3, like RealAudio and Windows Media Audio, belong to a class of audio compression formats (a/k/a “codecs”) known as “lossy.” That means audio information is discarded when the file or stream is created. The neat thing about lossy codecs is that they’re clever about what info they throw away. The strategy is to get rid of frequencies and information that are less likely to be heard or important to the overall intelligibility of the audio. In general, they work remarkably, but not perfectly. While some argue that CDs are also imperfect, pretty much every file encoded with a lossy codec, like MP3, is inferior to the CD original, though the differences range from nearly imperceptible to glaringly obvious.
Now with both home and mobile broadband, downloading MP3 files is trivial, requiring mere minutes to get a whole albums’ worth of music, rather than the hour or two it took over dial-up. Furthermore, there are much superior alternatives. That’s why I’m ready to bid the MP3 farewell.
The Right Tech at the Right Time, but that Time Has Passed
Don’t get me wrong, I have very warm feelings for the MP3 format. The eventual standardization on MP3 helped make internet radio nearly ubiquitous, especially by unchaining listeners from having to install two or three different proprietary players–I’m looking at you Real Player and Windows Media Player–in order to hear different stations. For broadcasters, MP3 still keeps bandwidth bills under control.
Moreover, podcasting owes its very existence to MP3, which still reigns as the medium’s default format, even if there are plenty of podcasters who pine away for something better.
The reasons for MP3’s popularity are also the reasons why it’s ready for retirement. MP3 was a brilliant solution at just the right time. At the turn of the century the format brought the ability to store and share a full CD’s worth of music in about 1/10 the space (at 128 kbps) at the same time that computers became powerful enough to play these files effortlessly and hard drives became big enough–holding as much 100 GB, or 1,500 CDs in MP3–to amass a sizable music collection. Though far from sonically perfect, MP3 was better than anything else you would find online.
Now, even MP3s have improved over the last fifteen years as the compression and playback software have gotten better. Also, the distribution of music has moved to higher bitrates (256 kbps – 320 kbps) that deliver substantial sonic benefits while maintaining bandwidth efficiency.
At the same MP3s have been outclassed by newer compression technologies–like AAC (Advanced Audio Coding) and Ogg Vorbis. While there is a strong case for internet radio to continue using lossy codecs, MP3 is no longer the best choice.
When it comes to selling music downloads, I also think that it’s unnecessary to use MP3s. Even other lossy compression formats (including AAC) no increasingly less necessary. At the very least they should be only one of many available options. I’ll take up this aspect in a follow-up post.
MP3 No Longer Needed for Internet Radio
Using lossy compression for streaming music and internet radio is still both reasonable and practical because there are good reasons for both broadcasters and listeners to want or need to conserve bandwidth. For broadcasters, it costs money to serve each listener, and so the difference between delivering 128 kbps and 1 mbps per stream can add up quickly when we’re talking about thousands of listeners.
Listeners on mobile may only be able to receive so much bandwidth, and they’re much more likely to be paying for every megabyte, or have a monthly quota. At home the case for lossy compression is less clear-cut, especially when people routinely watch streaming video that consumes several times the bandwidth. But I’m also realistic, recognizing that users listening through tinny PC speakers or cheap earbuds have little for full CD quality. Plus, a home with a few teenagers in it can quickly test the bandwidth limits of your cable modem service.
But not needing full CD quality is not an argument for MP3. Alternative formats like AAC or the open source Ogg Vorbis deliver a better experience at equivalent bitrates. Importantly for mobile users, variants of AAC provide surprisingly good fidelity over very restricted bandwidth conditions where MP3 streams sound pretty atrocious.
For the most part, internet radio listeners are ready to migrate away from MP3, too. Most desktop player apps–like iTunes or WinAmp–handle formats like AAC. Plus, many desktop listeners use embedded web players that can handle other formats, too. So do the Android and iOS mobile operating systems, meaning some 96% of mobile users in the US are ready for a switch.
Pretty much the only listeners not ready for the switch are those using older dedicated internet radios that either don’t have the capability to use other formats, or that don’t support the protocols used to deliver them. That’s the problem with the BBC’s recent near-wholesale transition away from MP3, which has left some internet radio users less well served. However, from both a cost and technical standpoint, in most cases it’s not a big deal to leave a legacy MP3 stream running.
Better Alternatives for Internet Radio
Realistically, while analog radio analog FM radio can sound very good, broadcasts only occasionally approach fidelity one might call CD-quality. So, I’m perfectly happy with internet radio that achieves similar results. I’ll even admit that high bitrate MP3 streaming (256kbps and higher) can meet that standard.
The thing is, AAC definitely qualifies as near-CD quality at 256 kbps, and I’ve heard AAC+ streams at bitrates as low as 96 kbps that meet or beat 256kbps MP3 streams (Radio Swiss Jazz’s 96kbps AAC+ stream is one example). That means using AAC offers potential cost savings for stations in terms of bandwidth, while also offering higher fidelity to listeners. I’d call that a win-win.
On top of AAC’s greater bandwidth efficiency, using it actually costs a broadcaster less than MP3. That’s because running an MP3 stream requires a royalty fee that many people–and broadcasters–are probably not aware of. Those fees start at $2,000 a year, though non-commercial and non-revenue-generating broadcasters should be exempt. Although MP3 is freely used and feels free to average user, it still contains technologies that are licensed under patent.
Now, AAC has patent-protected technology, too. The difference is that it is explicitly licensed for broadcasters to use for free.
Why MP3 Persists
Given all of this, the only reason why MP3 still seems to dominate internet radio is just tradition, really. The support for AAC, in particular, is nearly as widespread as MP3, and it costs less. Pretty much all MP3 has going for it is name recognition and the fact that many stations otherwise don’t have the incentive to fix what does not appear to be broken.
Surfing around stations I certainly find many that do offer AAC streams. My completely unscientific sample shows that amongst terrestrial broadcasters public and community stations are more likely to offer an AAC stream, or even make AAC the default format, as Seattle’s KEXP does. Commercial stations often only provide an in-browser player where the format is unknown, or direct you to a platform like iHeartRadio, which is similarly opaque (though I suspect iHeartRadio actually uses AAC for its mobile apps).
Pure-play internet stations are all over the map. However–again, unscientifically–it seems like stations with a web presence that are music-focused are more likely to offer not just AAC, but a variety of different codecs and bitrates. San Francisco-based Soma FM is one example. This listener-supported platform of 28 different stations offers at least three bitrates each of MP3 and AAC.
To be clear, I don’t necessarily care if a station explicitly tells me what format it streams in, as long as that stream sounds good. It’s just that too much internet radio doesn’t sound that great, and when I dig under the hood I usually find an MP3 stream (and usually a lower bitrate one, at that). I don’t blame a broadcaster for wanting to save money with a lower bitrate–I only wish they’d use a better codec than MP3.
What About Full CD-Quality Internet Radio?
Given my preference for full CD-quality digital music that doesn’t use lossy compression, I’m also an advocate for its use in internet radio. I’ve even written about the few stations that deliver these streams.
However, I’m also aware that streaming CD-quality FLAC or ALAC means 5 to 10 times more bandwidth for a station to cover, at a commensurately higher cost. Additionally, support for lossless streaming is much less universal than it is for AAC. For instance, iTunes doesn’t support FLAC, while pretty much only Apple products support ALAC natively. A station can offer an in-browser player that will work on most desktops, but that still counts out a healthy percentage of potential audience.
For the same reasons why I’m OK with analog FM’s near CD-quality fidelity, I’m fine listening to a good internet stream that matches the sound quality of a decent FM broadcast. Sure, I’d love to hear more uncompressed internet radio, but I can live without it.
Streaming music services are a different deal. First, I’m pretty sure the major streaming services have all abandoned MP3. As one would expect, Apple-owned Beats and iTunes Radio both use AAC, as do Rdio and Pandora. Rhapsody has used both AAC and MP3 in the past, while Spotify uses Ogg Vorbis. Google Play Music and Amazon Prime Music are stuck with MP3. That said, the paid subscription versions of these services generally sound pretty good under most circumstances–paid subscribers, predictably, get served the highest bitrates. I think they’re a reasonable value, but I’m personally not terrifically happy to pay for access to lossy streams.
I’m glad to see the recent introduction of full CD-quality streaming services like TIDAL and Deezer Elite, and I’m perfectly good with paying the subscription fee to get them. I also have no problem with the fact that CD-quality costs more than the lossy versions because I do hear and appreciate the difference.
The War of Attrition
The good news is that I do think AAC is edging MP3 out, primarily due to its superior sound quality over bandwidth restricted mobile broadband. For big streaming platforms using AAC means both savings and preventing mobile listeners from choosing other apps that sound better.
Yet, there are still hundreds, if not thousands, of internet stations–commercial and noncommercial–that still only offer MP3, often with middling to poor sound quality. I don’t want to call any out, but I’m surprised how many great community and college stations I like still offer up mediocre-sounding MP3 streams. They could sound better without any additional cost–maybe even some savings.
It may seem too geekily technical or esoteric, but I think internet radio can and should sound good, and that radio stations–especially great independent stations–will only thrive by embracing new, better sounding technologies.