Originally Posted by
C14ru5
There are a lot of myths and personal opinions when it comes to lossy vs. lossless compression formats, and you'll probably get 100 different answers. Here's my view:
1. In a blind test, you may always hear a difference, but that doesn't necessarily mean that you can sort the lossy compression from the control (the unmodified test sample) ;) But that's the endless debate of hearing the difference. There is always a technical difference, which can be seen if you compare waveforms. When it comes to digitizing music that you are to process later, you'll want to keep a lossless format for as long as possible until you're done with the audio processing. Apart from a few very special audio codecs through history (like BBC MPEG1 Layer 2), re-compressing an audio file will degrade the signal a bit more every time you do so.
The answer to the second half of question 1 - "Is [hearing the pre/post compression difference] relevant when the source audio is a vinyl rip rather than a professionally produced CD?":
- I would argue that it is actually more relevant when ripping vinyl than a CD. The challenges for the compression algorithms are the same, they're just more abundant in the case of vinyl. Let me explain: The MP3 algorithm uses time-window analysis to determine the ear's natural masking in three domains - frequency, time and also phase (for mp3s encoded in "joint stereo"). While fine nuances in frequency and phase may not be too ugly after compression (like diffuse overtone structures or out-of-phase effects), the by far most problematic variable is small nuances in time. A classic example is a simple transient, which is only 1 sample long, but through compression gets blurred out into the full length of the time-window. A vinyl record is full of such transients (pops and crackles), so the noise of the vinyl medium puts more stress on the compression algorithm than a typical "noiseless CD".
2. I know friends who spend lots of time using de-hissers and de-cracklers, and high-end software can sometimes do a surprisingly good job. I generally avoid using such tools myself, since you usually can hear artifacts from the process anyway, and the artifacts sound more disturbing than the original noise IMO. However, I do remove the worst popcorn effects manually, using the pencil tool in Audacity (one may just as well run a heavy low-pass filter across the few samples in question to get a similar effect).
Second half of question 2: Apart from most encoders using a 20Hz high-pass filter before encoding, I don't see any reason at all why compressing an audio file can produce a (technically) better sound. Unless we're talking the effects of dynamic compression, but that's a totally unrelated topic and is something that we should keep out of this discussion to avoid confusion.