bitrate quality

[sev]

our station broadcasts at 64K & 24K mono. In the forum users opinion do you think mp3s converted at 128K would sound good enough especially on the 64K stream or would there be a significant advantage of converting all our tracks again at a higher bitrate (quite a lengthy process)?

[djSpinnerCee]

When you upsample (increase bitrate) the quality will never improve. You would have to re-master the MP3 from the source, whatever it was.

It would be a better idea to downsample all of the tracks to the target bitrate to save processing by the DSP, but that's really not necessary either. From a "technical" standpoint, the ideal source bitrate would equal the stream bitrate. Some people also believe that even multiples help downsample quality as well -- 64kbps being half 128kbps, 32kbps = one quarter 128kbps, etc...

In my opinion, were you to have all 320kbps MP3s that you were forcing your DSP to downsample to 64 and 24 kbps, you would probably find poor quality because the difference is significant enough that you may introduce artifacts (=noise) and you will "lose" a lot more quality in the conversion that if you had 128 kbps MP3s. To put that another way, to your ear, the stream would "sound" far worse than the source.

Also, keep in mind that 64kbps stereo is actually 32kbps per channel, so they "sound" essentially the same. Stereo depth however, is sometimes more desireable than frequency response.

[sev]

Thats a good helpful answer,thanks matey. I think ill keep my mp3s at 128K as my stream is 64K (64 being half of 128)as some people say it helps downsample quality. Its just that when i tune in myself it sounds good,just like 128K but in mono and other times i notice artifacts more

[gaekwad2]

You're not downsampling (that would be going from 44.1 to eg. 32KHz), you're recompressing which always hurts quality.

If you're using sHOUTcast dsp your source always gets decompressed and reencoded even if it's already at the target bitrate (I don't know about other streaming progs).

In theory you should always use the highest source quality possible, but the difference between 64kbps transcoded from 128 or from high quality or even lossless probably isn't big enough to justify reripping your whole library.

@djSpinnerCee 320kbps would give better quality, reencoding in multiple steps is a very bad idea, and even if you're resampling going to 1/2 the original sampling frequency doesn't produce better results (resampler don't just drop samples, that would sound just horrible).