Hello,
Yesterday I used Cdex to encode a song by using –alt preset standard (lame 3.92). I got a song with an average bit rate of 159kbps. I used EncSpot to see how the bit reservoir was performing. So I found out that whenever 128kbps was used the bit reservoir was going up till 500.

Therefore, I encoded the same song again with –alt preset standard; however, this time I set the minimum bit rate to 96kbps. This time the bit reservoir looked ok – no extraordinary big or small values as before; I’d say a typical vbr encoded file. The strange thing is that the average bit rate was still the same as in the previous mp3 (159kbps). This is sort of interesting because it actually had 96kbps and 112kbps blocks.

Therefore, I did a third test. The same procedure as in the other 2 encodings, except that I set this time the minimum bit rate to 32kbps. Again, the bit reservoir looked like alright. But guess what, the average bit rate was again 159kbps.

So how is that possible? Setting the minimum bit rate to 128kbps was obviously a bit of a waste, since the bit reservoir had high values on some of the 128kbps encoded passages. So when I set it to 96kbps minimum I would have expected that the average would be a little lower though. I mean what were the bits used for I saved on this passages that were before 128kbps and are now 96kbps (or like in the third example even lower)?
Which file should I keep?

Also, the lame tag always stated the minimum bit rate as ‘unknown’ :confused:

I’m looking forward to read your comments on that

thanks

After the 128kbps passages the bit reservoir is emptied by using bits. This way e.g. a frame that would be 192kbps without reservoir is 160kbps using bits from the reservoir.


When the --alt-presets were created -b 128 was chosen because lower -b values decreased quality in some cases. So just trust the --alt-presets.

This is because when using -b 128, the unused space was not wasted, but capitalized. It was converted as bit reservoir space, which was used by following frames.

When using -b 96, the same following frames were probably bigger themselves, but using the same number of bits: first case big reservoir and small frame, second case smaller reservoir but bigger frame.

As your overall bitrate is the same, it means that in this particular track each easy to encode part is followed by an harder to encode part.


This is a bug of the lame tag when using presets with -b xxx.

thank you guys for your fast reply's . I don't know about the -b switch. I just used Cdex. And when I selected --aps it automatically jumped to 128kbps as the minimum bit rate. So I assume that Cdex uses the b switch.
Anyway, thanks a lot for your advices and I'll keep the original --aps standard file