MP3 Bit Rate

Status
Not open for further replies.

Deziel0495

PEI/NB Database Guy
Database Admin
Joined
Feb 11, 2008
Messages
711
Location
PEI, Canada
Hello Everyone

I know the RR standard bit rate for mono feeds is 16 kbps and 32 kbps for stereo feeds.

Just wondering if anyone is feeding at a higher bit rate or lower bit rate than the RR standard.
I find my mono feed is really "pixelated" at 16 kbps.

I've seen a few feeds with bit rates as high as 128 kbps. Can't remember what they were though.

I know the higher the bit rate, the more bandwidth it will use, but the better/clearer it sounds.

Thoughts, opinions, suggestion.... :)
 

longreach

Member
Feed Provider
Joined
Nov 30, 2007
Messages
143
Location
Southern Tablelands NSW Australia
Im running my personal feed(same feed as the RR one) at 32kbps sample rate 22050hz. It is a WMA feed though but sounds better. I wish i could run my RR feed at 22kbps, however i understand the issues why and just put up with it. if im away i quite often listen to the WMA feed, however if im listening thru my mobile phone its back to the RR feed
cheers
 

n5usr

Member
Joined
Aug 10, 2007
Messages
595
Location
Bethany, OK
I am also using 24k and 22050 mono. I played with all sorts of settings when I first set up my feed, and this was the lowest-bitrate setting I could find that still sounded good. (And that was compatible with most/all players.)
 

PeterGV

K1PGV
Joined
Jul 10, 2006
Messages
754
Location
Mont Vernon, NH
I agree that 24Kbps sounds better than 16Kbps, especially for most digital audio. It doesn't make a great deal of sense to me technically, but after my own experimentation I settled on 24Kbps bit rate for my own, private (non-RR) monitoring.

The 22.050Kbps sample rate is important for a lot of reasons, not the least of which is compatibility. Some players, notably certain Flash players, don't work or don't work correctly with sample rates other than 22.050Kbps. Therefore, ScannerCast *always* samples at 22.050Kbps, even with a bitrate of 16Kbps. There's a great deal of debate in the audio community as to whether this is a good idea for low bit rate audio, but (since I can change the code) I've tested both sample rates and have found the audio to certainly no worse (and probably a bit better) with a sample rate of 22.050Kbps.

In case you're not aware, while ScannerCast RR Edition fixes the bit rate at 16Kbps you can use the Standard Edition of ScannerCast (downloadable from my web site) to set the bit rate to whatever you like.

I hasten to add that I have no idea what RR's policy is about providing streams at bit rates other than 16Kbps...

Peter
K1PGV
 

GTR8000

NY/NJ Database Guy
Database Admin
Joined
Oct 4, 2007
Messages
10,869
Location
BEE00
Anything less than 22 kHz sampling definitely negatively affects the audio quality, as we found out a few months ago in ScannerCast! Besides, the file size/bandwidth remains the same regardless of the sample rate, it's the bit rate that affects file size/bandwidth. There is really no reason to use anything less than 22 kHz sampling, even at 16 kbit/s bit rate.

RR doesn't have a policy regarding bit rate, at least not that I've ever seen. 16 kbit/s seems to be their "recommended" bit rate, as I guess they feel it provides the best balance between decent audio versus required bandwidth. Personally, I don't agree. I find anything below 24 kbit/s to be less than ideal and having a noticeable degradation of the audio.
 

blantonl

Founder and CEO
Staff member
Joined
Dec 9, 2000
Messages
9,962
Location
San Antonio, TX
Our standard recommended bitrate is 16kbs, and we do not want feed providers providing feeds at other bitrates.

16 Kbs is more than acceptable standard for mono speech. Now, I know that everyone has an opinion about bitrates and "pixalizations" all other types of audio quality, but it works well and provides a good quality for feeds.

In addition, increasing the bitrates will dramatically increase the bandwidth costs associated with running the service. A 30% increase in bandwidth costs would result in tremendous amount of increased bandwidth costs.
 

GTR8000

NY/NJ Database Guy
Database Admin
Joined
Oct 4, 2007
Messages
10,869
Location
BEE00
16 Kbs is more than acceptable standard for mono speech. Now, I know that everyone has an opinion about bitrates and "pixalizations" all other types of audio quality, but it works well and provides a good quality for feeds.
I'm not quite sure how something quantifiable like bit rates can be considered "opinion", but I'm not looking to start a whole debate about it either. Suffice to say, my sensitive ears can very easily distinguish between different bit rates, and even the relatively small increase of 8 kbit/s between 16 and 24 makes a noticeable difference. 16 kbit/s is the lowest limit of acceptability for radio transmissions, it's certainly not "more than acceptable". A telephone line is equivalent to 8 kbit/s, AM radio is equivalent to 32 kbit/s, and FM radio is equivalent to 96 kbit/s, just to put it in perspective.

I understand you want to keep the bandwidth usage as low as possible, however our listeners are quite used to the 24 kbit/s quality we've been running on ShoutCast for over 4 years now. As well, we use that feed for archival purposes above and beyond the radio room loggers, because it's much easier to create audio clips of specific incidents/transmissions from the mp3 files we rip locally (unrelated to the RR archives). Since mp3 is already a lossy format, it's important to keep the bit rate high enough that re-compression post-editing won't degrade the clip to the point of it being unusable.
 
Last edited:

Tweekerbob

Member
Joined
May 27, 2004
Messages
617
I won't say the bit rate makes a difference, but in the end it is the sampling rate that makes the difference in terms of how "rich" the audio is. I will say that a 16 kpbs stream with a 11025 Hz sampling rate is sufficient to reproduce the human voice (ok, maybe not an opera singer).

We are not rebroadcasting music so I see no point in mentioning AM/FM radio (predominantly music).

I run a 16 kpbs, 11025 Hz feed with no degradation in audio quality. 5 KHz is more than enough to capture all of the fundamental and 99% of all harmonic frequencies (enough to be audible) of the human voice. Going to 10 KHz with a 22050 Hz sampling rate, gives you practically nothing more than the 11025 Hz sampling rate does. What it can do at times is cause the pixelation effect others have noticed. You are basically cramming in twice the frequency info into the same bitrate stream. The MP3 encoder needs to compensate for this, so you may hear artifacts of that process.

There are better voice codecs being used all over the world than MP3, but none is more universal. It's lossy by nature.

What editing techniques are being used that you need to "re-compress" already compressed audio?
 
Status
Not open for further replies.
Top