by Adveser » Sun Jul 05, 2009 2:05 am
I would be weary of using VBR at all if you use DSP's. In my experience, DSP's sound a bit different processing a 320KBPS file verses 160. I'm not claiming this is a fact, but it seems the DSP's have more effect the lower the bitrate. I don't seem to get any of this degraded quality business either from re-encoding, probably because I only re-encode to change sample rate from 44.1 to 48. Of course I use a standard bit-rate because of what I previously mentioned. The smart money says that a good encoder is not going to reduce the quality of a file going from 160K 44.1 to 160K 48, it's going to run it through it's algorithms and alter the waveform to match a 48K sample rate. According to the LAME documentation, the program essentially sends the information through a DSP which is capable of making decisions based on the information present as far as what to delete and what to modify for the better. if that is indeed the case, then it is going to use the extra bandwidth of the higher sample rate to allocate information by forcing some of the original into the expanded bandwidth. If you don't think this is true, load a 48K Lame encoded file and send it through the audiospacial DSP stacker and click the button "force 44.1" (even though it says so, it is NOT reccomended to use that setting by the way for it's intended purpose) The effect seems to indicate that a 48K lame mp3 is not a padded 44.1 file, but a different waveform. Depending on your attitude, that may be an improvement or degredation. Personally, if the encoding is not using any high/low/band pass filtering and is using a q0 setting and is maintaining true stereo, i don't see how it could be doing much harm. This is because it already filtered out the crap the first time, it isn't going to that every time. This is not analog technology. Re-recording a wave-form over and over in digital will not result in a loss, especially when you aren't dealing with actual recording problems. I'm not trying to argue this point really, it's basically subjective, and what we know about subjective audio quality is that it is not a reliable indicator of quality, just perception. What i'm saying is that if you upconverted a lossy file to one of the same type, the encoder is simply going to pad the new file with redundant information to maintain the integrity of the original because it doesn't have the bigger file to make the decisions on what to transfer.
I'm no expert in how encoders work, but I read that 100 page manual that came with lame.exe and after working with the format for about 8 years, these are my observations.
I would be weary of using VBR at all if you use DSP's. In my experience, DSP's sound a bit different processing a 320KBPS file verses 160. I'm not claiming this is a fact, but it seems the DSP's have more effect the lower the bitrate. I don't seem to get any of this degraded quality business either from re-encoding, probably because I only re-encode to change sample rate from 44.1 to 48. Of course I use a standard bit-rate because of what I previously mentioned. The smart money says that a good encoder is not going to reduce the quality of a file going from 160K 44.1 to 160K 48, it's going to run it through it's algorithms and alter the waveform to match a 48K sample rate. According to the LAME documentation, the program essentially sends the information through a DSP which is capable of making decisions based on the information present as far as what to delete and what to modify for the better. if that is indeed the case, then it is going to use the extra bandwidth of the higher sample rate to allocate information by forcing some of the original into the expanded bandwidth. If you don't think this is true, load a 48K Lame encoded file and send it through the audiospacial DSP stacker and click the button "force 44.1" (even though it says so, it is NOT reccomended to use that setting by the way for it's intended purpose) The effect seems to indicate that a 48K lame mp3 is not a padded 44.1 file, but a different waveform. Depending on your attitude, that may be an improvement or degredation. Personally, if the encoding is not using any high/low/band pass filtering and is using a q0 setting and is maintaining true stereo, i don't see how it could be doing much harm. This is because it already filtered out the crap the first time, it isn't going to that every time. This is not analog technology. Re-recording a wave-form over and over in digital will not result in a loss, especially when you aren't dealing with actual recording problems. I'm not trying to argue this point really, it's basically subjective, and what we know about subjective audio quality is that it is not a reliable indicator of quality, just perception. What i'm saying is that if you upconverted a lossy file to one of the same type, the encoder is simply going to pad the new file with redundant information to maintain the integrity of the original because it doesn't have the bigger file to make the decisions on what to transfer.
I'm no expert in how encoders work, but I read that 100 page manual that came with lame.exe and after working with the format for about 8 years, these are my observations.