hi all
ive been getting mp3 files in 128 160 192 bitrates and then decoding them to wav files using RazorLame 1.1.5 and lame 3.96.1 and then encoding to mp3 again using -b 128 -m s -h
i do this coz i prefer the 128 bitrate.
just wondering what is everyones opinion on this, how iam doing it , the programs iam using ect.
and pls remember iam new so pls explain clearly, thanks.
Angelus
Full Version: mp3 to wav to mp3
evry time you encode to mp3 you will reduce quality
so when you take a 192kbits mp3 and further compress it to 128 the qualtiy will be far below a "virgin" 128kbits encoding.
in other words
you are making a hell of a bad mp3 file.
so when you take a 192kbits mp3 and further compress it to 128 the qualtiy will be far below a "virgin" 128kbits encoding.
in other words
you are making a hell of a bad mp3 file.
QUOTE(Angelus @ Feb 8 2005, 01:07 PM)
hi all
ive been getting mp3 files in 128 160 192 bitrates and then decoding them to wav files using RazorLame 1.1.5 and lame 3.96.1 and then encoding to mp3 again using -b 128 -m s -h
i do this coz i prefer the 128 bitrate.
just wondering what is everyones opinion on this, how iam doing it , the programs iam using ect.
and pls remember iam new so pls explain clearly, thanks.
Angelus
ive been getting mp3 files in 128 160 192 bitrates and then decoding them to wav files using RazorLame 1.1.5 and lame 3.96.1 and then encoding to mp3 again using -b 128 -m s -h
i do this coz i prefer the 128 bitrate.
just wondering what is everyones opinion on this, how iam doing it , the programs iam using ect.
and pls remember iam new so pls explain clearly, thanks.
Angelus
Reencoding like this, better known as transcoding always reduces the quality of your mp3s. Stop doing it. Also, it is best not to use "custom" command lines. --preset cbr 128 will sound better then your line. If you don't mind using VBR "-V5" is even better. For LAME 3.96.1 you can improve the quality even farther by adding --athaa-sensitivity 1, but this is only due to a bug in this version and will probably reduce the quality in future versions of LAME.
edit: forgot to close a quote.
I do this to make files for my mp3 player.
Theoretically there are definite sympathetic/unproblematic aspects to transcoding
along side the potential hazards such as reported complications that might arise
from eg repeated low pass filtering.
fwiw I theorise that as long as often mentioned, but rarely defined potential double processing side effects are avoided, and a sensible preset is specified, there may be little extra degradation in sound quality from transcoding down than the reduction in bitrate itself implies.
Your commandline of -b 128 -m s -h
could cause problems because -m s -can require much more
bits than the default -m j , joint stereo, which is reported to be solidly transparent
and is highly recommended, so by disallowing lame to use this mode when approprate it will likely choke the available bitrate leading to audible artifacts in difficult parts of the songs.
I get nice results with using foobars automated conversion component.
using the high quality polyphase resampler to convert to 32Khz and send
that to lame using -b128 --cbr. (folks liking 128cbr mostly dont mind 32Khz bw ; )
The resampling hopefully takes care of any problem with repeated lowpass,
or -k could possibly be used to disable bandwidth filtering altogether, but it sounds ok to me and there are reported benefits of having a transitional end to the frequency spectrum, which lame uses for 128 cbr.
Feeding lame well resampled 32k audio should allow it bitrate headroom when encoding to 128 cbr as lame doesnt automicaly resample to 32k until <100 kbs.
The goal is to not reinforce any existing artifacts by making sure its not trying to squeeze too much bandwidth into too little bitrate.
It might be advantageous to turn off other effects such as temporal masking.
but for each effect disabled more bitrate will be required to avoid artifacts.
hth
The results of transcoding using lame -b128 --cbr using this method are fine to my nonaudiphile ears and gear.
Theoretically there are definite sympathetic/unproblematic aspects to transcoding
along side the potential hazards such as reported complications that might arise
from eg repeated low pass filtering.
fwiw I theorise that as long as often mentioned, but rarely defined potential double processing side effects are avoided, and a sensible preset is specified, there may be little extra degradation in sound quality from transcoding down than the reduction in bitrate itself implies.
Your commandline of -b 128 -m s -h
could cause problems because -m s -can require much more
bits than the default -m j , joint stereo, which is reported to be solidly transparent
and is highly recommended, so by disallowing lame to use this mode when approprate it will likely choke the available bitrate leading to audible artifacts in difficult parts of the songs.
I get nice results with using foobars automated conversion component.
using the high quality polyphase resampler to convert to 32Khz and send
that to lame using -b128 --cbr. (folks liking 128cbr mostly dont mind 32Khz bw ; )
The resampling hopefully takes care of any problem with repeated lowpass,
or -k could possibly be used to disable bandwidth filtering altogether, but it sounds ok to me and there are reported benefits of having a transitional end to the frequency spectrum, which lame uses for 128 cbr.
Feeding lame well resampled 32k audio should allow it bitrate headroom when encoding to 128 cbr as lame doesnt automicaly resample to 32k until <100 kbs.
The goal is to not reinforce any existing artifacts by making sure its not trying to squeeze too much bandwidth into too little bitrate.
It might be advantageous to turn off other effects such as temporal masking.
but for each effect disabled more bitrate will be required to avoid artifacts.
hth
The results of transcoding using lame -b128 --cbr using this method are fine to my nonaudiphile ears and gear.
QUOTE(ChiGung @ Feb 14 2005, 12:52 PM)
I do this to make files for my mp3 player.
Theoretically there are definite sympathetic/unproblematic aspects to transcoding
along side the potential hazards such as reported complications that might arise
from eg repeated low pass filtering.
fwiw I theorise that as long as often mentioned, but rarely defined potential double processing side effects are avoided, and a sensible preset is specified, there may be little extra degradation in sound quality from transcoding down than the reduction in bitrate itself implies.
Your commandline of -b 128 -m s -h
could cause problems because -m s -can require much more
bits than the default -m j , joint stereo, which is reported to be solidly transparent
and is highly recommended, so by disallowing lame to use this mode when approprate it will likely choke the available bitrate leading to audible artifacts in difficult parts of the songs.
I get nice results with using foobars automated conversion component.
using the high quality polyphase resampler to convert to 32Khz and send
that to lame using -b128 --cbr. (folks liking 128cbr mostly dont mind 32Khz bw ; )
The resampling hopefully takes care of any problem with repeated lowpass,
or -k could possibly be used to disable bandwidth filtering altogether, but it sounds ok to me and there are reported benefits of having a transitional end to the frequency spectrum, which lame uses for 128 cbr.
Feeding lame well resampled 32k audio should allow it bitrate headroom when encoding to 128 cbr as lame doesnt automicaly resample to 32k until <100 kbs.
The goal is to not reinforce any existing artifacts by making sure its not trying to squeeze too much bandwidth into too little bitrate.
It might be advantageous to turn off other effects such as temporal masking.
but for each effect disabled more bitrate will be required to avoid artifacts.
hth
The results of transcoding using lame -b128 --cbr using this method are fine to my nonaudiphile ears and gear.
Theoretically there are definite sympathetic/unproblematic aspects to transcoding
along side the potential hazards such as reported complications that might arise
from eg repeated low pass filtering.
fwiw I theorise that as long as often mentioned, but rarely defined potential double processing side effects are avoided, and a sensible preset is specified, there may be little extra degradation in sound quality from transcoding down than the reduction in bitrate itself implies.
Your commandline of -b 128 -m s -h
could cause problems because -m s -can require much more
bits than the default -m j , joint stereo, which is reported to be solidly transparent
and is highly recommended, so by disallowing lame to use this mode when approprate it will likely choke the available bitrate leading to audible artifacts in difficult parts of the songs.
I get nice results with using foobars automated conversion component.
using the high quality polyphase resampler to convert to 32Khz and send
that to lame using -b128 --cbr. (folks liking 128cbr mostly dont mind 32Khz bw ; )
The resampling hopefully takes care of any problem with repeated lowpass,
or -k could possibly be used to disable bandwidth filtering altogether, but it sounds ok to me and there are reported benefits of having a transitional end to the frequency spectrum, which lame uses for 128 cbr.
Feeding lame well resampled 32k audio should allow it bitrate headroom when encoding to 128 cbr as lame doesnt automicaly resample to 32k until <100 kbs.
The goal is to not reinforce any existing artifacts by making sure its not trying to squeeze too much bandwidth into too little bitrate.
It might be advantageous to turn off other effects such as temporal masking.
but for each effect disabled more bitrate will be required to avoid artifacts.
hth
The results of transcoding using lame -b128 --cbr using this method are fine to my nonaudiphile ears and gear.
I really think resampling isn't a good idea, and I think lowpass isn't the only thing that needs to be taken into account.
I'm pretty sure I'd be able to ABX 44.1kHz vs. 32kHz. It'd probably be non-transparent for me.
QUOTE(kjoonlee @ Feb 14 2005, 07:07 AM)
I really think resampling isn't a good idea, and I think lowpass isn't the only thing that needs to be taken into account.
Its not the definite disaster that it is proclaimed to be, it does seem to work if done in a suitable way. And since the aural results are not problematic for some people it can be of use, at least for certain personal mp3 players and ears.
QUOTE
I'm pretty sure I'd be able to ABX 44.1kHz vs. 32kHz. It'd probably be non-transparent for me.
You probably have decent ears, but miss the point of transcoding, its not to be transparent its to achieve a lower bitrate or change to cbr.
I just compared two versions of 'theres a moon in the sky (called the moon)' by b52s in my flash mp3 player, origional at 224 kbs and transcoded version at 128 kbs and was unable to tell which was which, even my guess turned out to be wrong when i checked the filenames on the computer, and if I could tell the diff - its not certain the difference would bother me when Im actualy listening to the *music* rather than the minutiae, so which version should I keep in the mp3 player? and why shouldnt I make more such files?
In other quality related topics abx data is commonly required to back up claims, try some testing or inform us of specific technicalities before writing this one off as foolhardy
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.