i found some people said that a mp3 stream which encoded to insane bitrate 320 then re-encoded to other lower bitrate (example: 128) have better quality than directly encoded to bitrate 128.

it's that true?
perhaps, you can make a simple digital analys

thx

No, that is completely incorrect.

Not only is it incorrect, it will most likely lower the quality significantly.

Call me kenshin, idiots that fool themselfs are everywhere
Or maybe he/ she likes it more with noise
The quality is always better from lossless source.

may i know, how do you compare it

ABC a directly to 128 kbit/s converted mp3 against one that was created from a 320 kbit/s mp3.
EDIT: Of course, both the first and the second file should be created from the same lossless source.

Not only that is not true, but actually the opposite is true.


Direct encode -> no initial loss of quality.
Reencode from a higher bitrate -> initial loss of quality in the first encode.

There is no fixed rule as to how much the quality will degrade, just because some songs require more bitrate for the same quality than others, so in some, the problems would be more apparent.


For all intents, lossy to lossy (mp3 320 -> mp3 128 in this case) is something that should be avoided, if other options ( lossless/original -> lossy ) are available.

I wrote a thread at Head-Fi once to demonstrate to people there that transcoding from a lower bitrate to a higher bitrate degrades quality:

http://www.head-fi.org/forums/showthread.php?t=172742

Either way you transcode, it will really degrade quality. What's more, you made a 7th generation mp3 file from 64kbps.

If the question is which file will sound most like the lossless original then I think we are all pretty much agreed that directly to low-bitrate will beat the version with intermediate high bitrate. Any deviations in the first conversion will be augmented and possibly compounded by the second conversion.

However, if the question is not which sounds most accurate but which sounds "best" then personal taste becomes involved and I think the answer is not so clear-cut.

Of course.



As I indicated in the first post of that thread, the purpose of that thread was to disprove the assertion that was being made at Head-Fi at the time that transcoding to higher bitrates improved quality. For example, one Head-Fi member had written: "Another testament that converting files to a higher bit rate does improve the quality. Over the last 3 years I've upgraded my whole music library from 66>132>256> and finally 352kbsps. Each time I upgraded there was subtle improvement in sound quality. If all it was was an increase in volume I'm sure I would've noticed..." Others had argued that while transcoding to a lower bitrate was bad, files transcoded to a higher bitrate would maintain the quality of the original file.

My goal was to demonstrate successive transcoding resulted in an unmistakable degradation in quality, even where one is transcoding to a higher bitrate. As I explained in that thread, the reason I made a 7th generation file is because the difference between the original lossy file and the first-generation transcoded file was relatively small, and the artifacts introduced by a single transcode can be subjectively interpreted as an improvement to the sound of an mp3. Doing the demonstration with 6 separate transcoding steps, however, made the file degradation obvious.

I thought that demonstration would be relevant here because I read the initial post as saying that taking an existing MP3, transcoding it to 320kbps and then again to 128kbps would result in a higher quality than transcoding it directly to 128kbps. In other words, I read the initial post as suggesting that transcoding an MP3 to 320kbps would improve the quality.

I kinda find it hard to believe how much bullshit there is on the internet when it comes to sound quality. Reminds me of the medieval times somehow when people believed total nonlogical nonsense. I could claim that if you put water from a small glass into a big glass and back again, you afterwards have more water than before.

My view on this: lossless -> 128kbps is probably better than lossless -> 320kbps -> 128kbps. However, there is no direct way to prove this, if you want to be sure you have to perform a listening test.

I agree completely that any transcoding can only harm quality or leave it unchanged, because you cannot recreate the information lost in the original encoding process. 128kbps -> 320kbps is just bogus. But the situation is different here, because there are two encoding processes now which are both supposed to throw away some information. The "transcoding hurts quality" argument is true, and suggests the quality of lossless -> 128kbps is probably better, but can't be used to prove it.

For example, imagine a broken encoder, which tries to include all frequencies without lowpass or consulting a psymodel when encoding to 128kbps for some reason (I'm not well informed about mp3 encoding, so this example may contain errors). Now, encoding lossless->128kbps with this encoder would produce an inferior result, because a lot of "inaudible" data has been kept. Encoding from lossless to 320kbps first, the encoder would maybe lowpass the file a bit, and throw away a lot of inaudible data, producing a nice-sounding file. Then, encoding from 320->128, the result might be better that lossless->128, because a lot of inaudible data has already been discarded at the lossless->320kbps stage, and more bits are spent on getting the audible parts right.

I know this is a very constructed scenario, but I just wanted to make the point that, in this case, you can't be sure until you actually test.

Great ! I'll try that with my beer tonight.

Why'd you want a more watery beer?

I remember a teacher saying to me in HS that in medieval times they also believed the recipe for rats was to throw a bunch of dirty clothes in a corner and wait a week.