The Belgain
Feb 13 2003, 12:06
This is using WMA9 to compress a file which is mainly voice, but has some music and random sound effects too (Simpsons episode). What sampling rate should I use? I've tried with 32kHz and it sounds pretty bad (the same old wma artifact we you always get), should I go down to 22 kHz? The source is a 224kbit 44kHz stereo mp2.
Please don't give me a "don't use WMA, it sucks response". I'm using it because it's the only thing you can use in WMV atm. Once MS bring out their vfw version of WM9, I'll use Vorbis...until then I can't.
SometimesWarrior
Feb 13 2003, 14:20
I would definitely try going down to 22KHz. That sampling rate still sounds quite good for speech. You could even go down to 16KHz, and the voice should sound pretty good, but I would first try 22KHz.
mithrandir
Feb 13 2003, 15:53
Using mono instead of stereo should give you a notable increase in quality (at 32kbps/32KHz sampling). Think if it as the classic "pick 2 out of 3" dilemma:
low bitrate
high sampling rate
multiple channels
You want a low bitrate so you can only choose between a high sampling rate or 2 channels. For something like a TV episode, I'd go for wide-bandwidth mono.
Dibrom
Feb 13 2003, 16:09
QUOTE(The Belgain @ Feb 13 2003 - 11:06 AM)
Please don't give me a "don't use WMA, it sucks response". I'm using it because it's the only thing you can use in WMV atm.
Why are you restricted to WMV?
PatchWorKs
Feb 13 2003, 16:47
I use 48 Kbps/32KHz/mono that sounds really good to me.
...anyway i suggest you to use Soundprobe' expander (with 'variable' parameters) before encoding.
Check out
my band's OGGs to evaluate the quality. Good listning !
The Belgain
Feb 13 2003, 17:44
Right, thanks for all the replies. I'll try 22kHz stereo and 32kHz mono and see what the results are like. I guess going to mono for a Simpsons episode would be fairly bearable.
As to why I'm limited to wmv....I guess I'm not really. The only other choice is RM9 (I'm trying to make a 150kbps file, and there are not any other viable alternatives, at least until Theora comes out). However Real Producer has major synch issues when ripping from MPEG-1 files (because the MPEG-1 files synch themselves in a weird way - often the audio and video part don't have the same length once extracted - and Real doesn't deal with this very nicely...I frankly can't be arsed to do an Avisynth script for each file when I could just bung it straight into WME).
In addition to this, the (video) quality is actually much better than I expected in wmv. The audio doesn't really seem up to par yet though.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please
click here.