Hammeredklavier
Active Member
- Joined
- Feb 25, 2019
- Messages
- 112
- Likes
- 127
If you can reliably ABX MP3 vs RBCD at 320 kbps then I'm impressed. My hearing's pretty good and I don't think I'd stand a chance, unless a carefully chosen killer sample was used. Certainly if recent versions of the LAME encoder or the one by Fraunhofer were used! Very early MP3 encoders could sound pretty rough at lower bit rates I'll grant you.I usually pass 'modern' ABX tests 320kbps or VBR vs RBCD but with some constraints
- on my "big" systems (I am not a headphone guy).
- actively paying attention to details.
- on the type of music I like.
- on good recordings.
- on recordings or configurations (say a chamber music quatuor) I am familiar with.
A lot of constraints indeed...
On the other hand, when I play random tracks from my NAS, where I have stored my music collection (mostly rips of my own CDs, a few hires downloads) I can always almost instantly recognize my old 320 kbps files from my lossless ones. My ripping evolved over time, I started when encoding to 128 kbps was actually taking much longer than listening to the CD and when storage was expensive, only digitized everything to 320 kpbs a bit later, then finally went for a lossless ripping marathon... And since I am a fundamentally lazy person, I never cleaned up my mess which means I have duplicate 320kpbs/RBCD of nearly my entire library.
It may be subjective, or they may be objective reasons I am not aware of in the encoding behavior of old encoders, but the giveaway is almost always the width and depth of the sound stage in the case of my library. That really jumps in my face. Sometimes artifacts I have trained myself to detect too, but that requires paying attention.
That leads me to think either the encoder or the parameters choice weren't optimal when I did the 320kbps pass. But that was, of course, a time where legit and non-legit encoders proliferated, where some commercial encoders simply stole encoding libraries from open source projects and vice-versa.
Ultimately, I don't really want to worry about how the 320 kbps stream was generated and stick to RBCD, which has the added advantage of having convinced me it was all I needed for biological and mathematical reasons. As far as the recording/mastering process is concerned, on the other hand, I am all for giving engineers all the data they need and more, if simply to avoid things like the dirty voice track introduced in Amir's example.
And, the consistent instinctive choice, in multi-channel situations, would be to give every channel RBCD bandwidth. One less thing to worry about even if I don't really know and haven't investigated if that is even necessary (probably not on the whole frequency range anyway).
I'm not saying LAME is transparent at @120 kbps variable, but I'd have to be paying attention and I'd probably find the sound quality perfectly acceptable at that. With storage being so cheap I honestly haven't tried it for a bit.