top of page
Search
  • tim0666

SESSION NOTES 9 “Streaming Audio” (part 1 crap sound rules ok)




Hi everyone, this month I want to look at “Streaming” audio and in particular the formats and technical requirements of some of the more popular streaming sites. Streaming is a means of delivering data in a constant stream to the consumer as opposed to downloading. With streaming you can access the data immediately whereas with downloads you have to wait for the download to complete. All streaming services “data compress” the material with varying degrees of quality.

In this article I want to explain how to get the best sounding results when using these platforms and to introduce you to the new EBU (European Broadcast Union) specifications regarding level requirements by the most popular streaming sites.


The largest problem encountered by people using streaming sites is the loudness difference between tracks. Listening to whole albums seems to be a thing of the past with the emphasis now on single tracks, usually in a users playlist. The trouble is that all of these tracks have an inherently different loudness. This means you have to constantly fiddle with your volume control. Nearly all streaming sites now employ a “normalisation” algorithm to compensate for this. Unfortunately CD’s mastered to extremely loud (and un-dynamic) levels are pushed down whilst quieter (more dynamic) tracks are raised up slightly.

Well, what are they “normalised” to? Let’s look at “Spotify”. Launched in 2008 this is the worlds most popular streaming site commanding 36% of all streamed audio content. Figures published in July 2019 state 232 million users!!! All Spotify audio is normalised to -14LUFS. LUFS stands for Loudness Units (referenced to) Full Scale. The full spec of LUFS is referenced in the EBU R128 recommendation for broadcast and streamed audio. Loudness is about perception. Loudness normalisation on streaming platforms evens out these perceived differences.


Audio engineers are now required to submit material that adheres to this specification. Submitting highly maximised audio will result in distortion once the site encodes the music for streaming. So you will need to invest in a LUFS meter for your audio systems. Incidentally a ”LUFS” is equivalent to 1 dB. Quieter audio will sound significantly better than over-loud audio once the streaming platform has normalised the tracks. The idea is that you can set your preferred listening volume on your device and leave it there.


With the exception of “TIDAL”, streaming sites will data compress your audio. Data compression is totally different to audio compression. Audio compression as I’m sure you already know reduces the dynamic range of music. Data compression reduces the number of bits in a file size. Data compression will affect the quality of your sound, sometimes significantly. With an MP3 the data reduction can be as much as 90%. A 3 minute song at 16bit/44.1Khz (CD quality) will be around 30 Mb in size. The equivalent MP3 will be around 3Mb. 90% of the data will vanish. This can sound pretty dreadful at the best of times. With streaming, the data rate can be as low as 96Kbps (kilobytes per second) and as high as 320Kbps. CD on the other hand has a data rate of 1.4Mbps.


Formats -: Along with MP3 there are many digital audio file formats used in the industry for streaming and distribution. With the exception of “FLAC” and “WAV” they are all “Lossy”. Lossy means that data will be lost in favour of smaller file size. MP3’s have bit rates from 96Kbps up to 320Kbps. There are 2 kinds of bit rates, Constant Bit Rate (CBR) which is self explanatory and Variable Bit Rate (VBR). Variable bit rate works thus. Very sparse instrumentation can be encoded at a lower bit rate and when the track becomes more instrumentally dense it will be encoded with a higher bit rate. The unwanted artefacts from this data compression are grim at best but then if you insist on listening on a portable telephone … who gives a fuck. AAC (Advanced Audio Coding) format has nothing to do with Apple Corp. it was developed by a few companies, among them Bell Labs, AT&T and Nokia. It too is lossy and is regarded as “slightly” better than MP3. In fact nearly all the lossy formats are “slightly” better than MP3.


How popular is streaming? Well last year 431 billion streams were logged in the USA alone!!!!


(Next Month we will meet Ogg Vorbis … beware).

45 views0 comments
bottom of page