Quality diff. between 128, 256, etc bitrate MP3?

MikeLV wrote on 6/20/2013, 4:26 PM
How should one decide what bitrate to render an MP3 as? Why choose 192 or 256, etc over 128kbps? The usage of these MP3s will be for a pro musician's backing tracks played from a phone or tablet into a PA system. Is there any audible difference between anything above 128kpbs? Or is there a method to determine based on the music what bitrate is ideal? Thanks!

Comments

MikeLV wrote on 6/20/2013, 4:44 PM
I found some info on the topic after googling around, debates, tests, etc. But nothing really definitive. Some DJs say there's a world of difference between 320 MP3 and lossless WAV. Depends more on what you're playing the music through. I suppose the best way is to encode the same song in a few different formats and play it through the system to see if there's any difference in the sound. My guess is that I won't be able to hear any difference between 192 and 320....
musicvid10 wrote on 6/20/2013, 9:23 PM
The biggest problem with using mp3 as a backing track is the high frequency dropoff -- it is a cliff, and will have a harder time punching through the live music.

"If" you go with mp3, say in an acoustic setting, and it is stereo, 192-256 Kbps is a minimum, and I suggest you use LAME encoder (CDex is a good frontend).
I see no reason not to go with WAV, at CD quality as a minimum.

Here's the mp3 freq curve at 128 Kbps. It's much better at 256 K.
MikeLV wrote on 6/20/2013, 9:54 PM
Well I encoded a bunch of them to 320kbps MP3 and I used whatever encoder Vegas uses. I played them from the tablet, to the PA system and they sounded pretty darn good to me. They sounded just as good as when they were MIDI files being played through the sound engine.
musicvid10 wrote on 6/20/2013, 10:54 PM
Oh, they're midi. I've done a lot of those for rehearsal and backing tracks, and frankly it doesn't matter much how you encode them. The synth samples and processing are the limiting factor here.
rraud wrote on 6/21/2013, 10:46 AM
I used to archive spoken word material for a client and used 320kbps, 44.1kHz, CBR, mono.
In my experiments, the audio would totally cancel out when summed out of phase with the original PCM file.
I used the LAME RC3 encoder which supported 320kbp mono encoding and other UI parameter adjustments and other advanced parameters via a preset script file.
Complex harmonic content (music) would probably not fare as good and the quality would be halved compromised encoding in stereo.
MikeLV wrote on 6/21/2013, 10:53 AM
they WERE midi. I captured them all in real time onto the timeline in Vegas by playing them through a roland sound module, into a firewire device and then into the computer.
musicvid10 wrote on 6/21/2013, 2:30 PM
I understand, Mike. The synth (sound module) is likely the limiting factor wrt sample freq range and bit depth, Q noise, and harmonic distortion, and not your downstream encoder. That being said, midi source can sound nice, although not exactly of stellar quality.
Geoff_Wood wrote on 6/23/2013, 5:35 PM
"In my experiments, the audio would totally cancel out when summed out of phase with the original PCM file."

!!!

The encoder must have been not encoding then, or was in error coding to a lossless format !!! Unless you mean something less than actually "totally".

geoff
Geoff_Wood wrote on 6/23/2013, 5:41 PM
I's say go FLAC or PCA ? Or WAV. Unless the player has capacity or format limitations. The bare minimum for vaguely acceptable MP3 to my mind is 192kbps, which is kind of sad as most of the mp3s out there are (or were) done at 128kbps which is crcappy.

For myself I have an ipod (my only Apple gadget), and only use lossless encoding.

geoff
rraud wrote on 6/24/2013, 9:29 AM
"The encoder must have been not encoding then, or was in error coding to a lossless format !!! Unless you mean something less than actually "totally" "
- Well Geoff, 'total' may not be actual dB for dB, but resulted in cancellation of more than 78dBFS as I recall the meter resolution. Both the original WAVE.. and the LAME encoded MP3 were placed on the Vegas timeline, sync'd (to the sample level ... or close to it) and the phase inverted on one
Take into consideration, this was spoken word material @320kbps CBR MONO, which would hypothetically be the equivalent quality of a stereo MP3 @ 640kbps. .If you like, I could post sample files and you can try it yo'self..
I'm not saying an MP3 music file should be archived to the MP3 format, nor would I recommend it.
Cheers, Rick
musicvid10 wrote on 6/24/2013, 12:26 PM
At 320 Kbps mono, Lame is pretty darn close to lossless.
I wouldn't be too surprised at rraud's results.
PeterDuke wrote on 5/26/2014, 8:06 PM
I tried cancelling some 320 MP3, both music and speech, against the wave original. Speech cancelled more than music, but the difference was not silence, even for my poor ears. The peaks on the level meters seemed to be about 20 dB lower for music and 30 dB lower for speech. At each speech word I tended to hear a burst of noise with some vocalisation.

Since the MP3 is delayed from the original, I had to compensate by hand. I estimate that the residual delay error would have been a fraction of a sample period.

Interesting