Audio quality levels & file formats

Streams published on our platform are made available in different quality levels (bitrates) and formats, based on the package.

Qualities

Different quality levels are employed to enable a better experience for users on slow (or expensive) internet connections while still allowing other users to receive high quality audio.

File formats

Our streams are available in different formats to ensure wide compatibility accross browsers, software and devices:

We recommend (and use as default) AAC for any devices that support it, which is almost all devices.

Notes on bitrate

Many other stream hosting companies may offer up to 320kbps streams, while our highest supported bitrate is 160kbps.

But this does not mean our feeds are "half the quality" - in fact the opposite, using lower bitrates may provide real benefits to users.

MP3 Background

When MP3 was released (back in 1991!) it revolutionised audio compression. The early MP3 encoders were not very efficient at all, and music had to be encoded at the maximum allowed bitrate in MP3 (which is 320kbps) in order to prevent noticable audio artefacts.

One of the strengths of MP3 was that it only stipulated how the outcome must be packaged, not how an encoder must work. This meant that while all MP3 files were fully interoperable between players, encoders could experiment with how to get the best compression results.

MP3 encoders has had 30 years to mature, and modern encoders produce results that simply does not need a bitrate of 320kbps to produce good audio.

How do we measure audio codecs ?

What is "good audio" ? Audio compression codecs can generally be evaluated based on what bitrate (size of the data) it needs to reach "transparency".

Hydrogenaudio: If a lossily compressed result is perceptually indistinguishable from the uncompressed input, then the compression can be declared to be transparent.

In other words, a compression codec is "transparent" when the audio it produces can't be distinguished from the original uncompressed audio. Different compression codecs reach transparency at different bitrates:

So what is the best bitrates to use ?

As seen above this is largely determined by the codec being used.

We default to AAC for all our streaming (and podcast) content, using MP3 only as a fallback option. AAC is transparent at 160kbps so that means our "High" quality level on Commercial packages already encode audio quality that can't be distinguished from the original audio.

In reality though, lower bitrates work completely fine for almost all listeners, for example Soundcloud used 128kbps MP3 for years to distribute audio and the bulk of users were happy with this.

Most radio streams globally uses either 128k MP3 or 64k AAC streams.

What other factors are there ?

When deciding on bitrate, two other factors are important

• What equipment will be used to listen to this? The overwhelming majority of users will be listening to radio streams on their phone speaker, computer speakers or home assistant devices and most of this will happen in an open space or noisy environments like the kitchen, office or car. In these scenarios, the limiting factor to quality will never be the codec, but rather the equipment and environment where the audio is consumed. Only a very small percentage of listeners will be using high-fidelity equipment in quiet spaces where codec quality may become an issue.
• What bandwidth are they using to listen? It varies by market, but it is a good assumption that 50% or more of radio stream content will be consumed by mobile phones, frequently on slow and/or expensive mobile networks. In these cases providing a smaller size (lower bitrate) stream can mean the difference between an affordable and reliable stream or one the user cannot access due to cost or bandwidth constraints.

Conclusion

Bitrates are important, but vastly overrated in the market as a factor behind quality. The more important questions to ask is how listeners will be consuming the audio, and what data size and quality trade-offs is important to them.