Occasional reader and blogger Robert has broken down all the information on audio bit rates and audio formats.
His work can be found here: Ultimate Guide to Audio Bit Rates and Formats
It is a good read, especially for those that use audio streaming as their main content distribution method.
Streaming-only stations used to be a big thing but have been supplanted by Spotify and Pandora. I am not a huge fan of either of those services but I do like to listen to podcasts.
Good audio should be near the top of the list for any content provider. Few things are more annoying than listening to an interesting podcast with low volume, background noise, or other technical defects.
Interesting new format.
Don, it is more optimized for mobile…
The article seems mostly solid, but the author does keep stating that 128 kb/s is 128 Kilo BYTES and that bitrate is actually “byterate” in his first section “What is audio bitrate”, which is incorrect.
Common error with non-IT folks, but it’s a big difference. kb or Kb (small “b”) generally means kilo (thousand) bits per second. kB or KB (large “b”) generally means kilo (thousand) BYTES per second. 8 bits in a byte, so he’s inflating the bitrate of a 128kb/s mp3 to almost the bitrate of a CD.
The basics to know are that you can break things down thusly:
Lossy vs. lossless – basically is your codec throwing away any data to achieve a reduction in file size (and bitrate)? Lossy is usually doing fancy math to throw away data, lossless compression is like the “zip” algorithm used for computer files – you can fully restore the original.
Bit depth – 16bit vs. 24bit vs. ??? – the takeaway is that 16bit is fine for most single-generation (ie: record something, save it, play it back) uses. If you know you’re going to be running a recording through a ton of different processes, mixing, mastering, etc. 24bit gives you more headroom.
Sampling rate – 44.1KHz is standard CD, 48KHz DAT, and even higher for various A/D converters. Questionable whether playing back dog whistles is necessary. Google “nyquist theorem” to see how these numbers are arrived at. Basically sampling rate is the limit on how high a frequency you can save/reproduce.
Also, Paul – I’m sure you’d take issue with comparing a 128Kb/s MP3 to “broadcast radio quality”. FM is one of the few remaining places where you can sometimes find an all-analog chain from studio to receiver. And even if you’re not an analog purist, a well-engineered FM station not compressing the hell out of things sounds sooooo much better than a 128kb/s mp3 stream. No contest, IMHO. 🙂
I listened to all the digital samples in the link a couple of times.
I personally drive a lot, and spend time listening to broadcast radio in the various cars, mostly all kinds of music. I do a lot of button pushing, scanning, tuning around, etc. When I get too far “out in the sticks” for FM, I do listen to AM as well.
Unfortunately, quite a few of the FM stations in the Albany NY market and small markets to the north sound EXACTLY like the 128k MP3. You can hear the vocals, hear the notes, and tell what instrument played them, but there is no rich quality to the recovered sound.
Switching to the 32k sample with its low-pass audio filter was just like switching from listening to a small AM station’s FM translator to its main AM signal as one drives away from the tower, with AM’s more limited frequency response.
To me, the cymbals in the samples on the link sounded digitized at all qualities. But I was listening on laptop speakers, not studio monitors. There is a bass guitar line present, mixed way below the lead guitar and organ. It is not noticeable at 128k unless you search for it, but is noticeable at CD quality. Interestingly, it stands out the most in the 32k sample with all the highs removed.
The worst broadcast audio I have heard on the air was at WYVS (FM) in Speculator. A couple of years ago, it sounded like the satellite audio stream had been cranked up to about 230% modulation, and then the “audio adjuster” (I won’t say engineer) set the limiter to “correct” this. The rapid “pumping” practically removed all the vocals from music, and the network newscast audio feeds also. It has since been re-adjusted. But their choice of music program service still leaves a lot to be desired.
On the other side of the coin, the BEST broadcast audio I have heard locally was on Townsquare’s “transmitter in search of a format” on 105.7. It is currently running the local “Q” format in place of the Glenmont and Stone Arabia transmitters.
When they were in their “Rewind” format days, old songs that I never liked (i.e. couldn’t stand) would come on the air. The dynamic range and audio quality was so good, it made me stop and listen to the arrangement, production, and instrumentation used behind the vocals. My vehicles became like concert halls, with no one in the house seats but me. I began to look forward to these songs instead of changing the channel. Switching back and forth with other stations in the market made the difference in audio quality quite evident.
My hat is off to whoever set up that audio chain. I would love to see how it had been done.
Interesting comment, Tom. Regarding 105.7 in Albany; I will pass along your comment to the guy that did that work, it will make him pleased.