Skip to main content
Topic: 'Quality' setting questions (any encoder) (Read 2392 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

'Quality' setting questions (any encoder)

What I always wondered about is this:

What was the designer of this system shooting for when zie made the settings up? Like for example: What's '10' supposed to have? Is it supposed to have no artefacts in it, therefore, that's why the bitrate is almost near (above 550 kbps using GT3b2 encoder) what some of my FLACs turn out to be (above 600 kbps)? What's expected to be heard (as in, artefacts) at 4? Why a '-1' or any negative option? Also, what gets taken away as you start lowering the quality settings beginning at 10?

Thanks...

'Quality' setting questions (any encoder)

Reply #1
Levels >7 are not really tuned (if I recall an old quote from Garf..) and just uses more bits. GT3 should be transparent from quality 6. Lower levels have a HF hiss or noise that becomes more apparent as you drop the bitrate. There are some reports that the new aotuv encoder is transparent at q5 or 4 + Hf noise is greatly reduced , obviously more testing is needed. Results from the 128k test should be interesting. Why -1 ? I don't know , maybe its saying something about the quality .
wavpack 4.8 -b3x6c

'Quality' setting questions (any encoder)

Reply #2
Quote
Why a '-1' or any negative option?

Because they added a new even-lower-quality setting after people had already gotten used to the original scale. There was some debate at the time about adjusting the scale to avoid negative numbers, but they decided that it would be less confusing to existing users if they didn't.

'Quality' setting questions (any encoder)

Reply #3
Quote
Levels >7 are not really tuned (if I recall an old quote from Garf..) and just uses more bits. GT3 should be transparent from quality 6. Lower levels have a HF hiss or noise that becomes more apparent as you drop the bitrate. There are some reports that the new aotuv encoder is transparent at q5 or 4 + Hf noise is greatly reduced , obviously more testing is needed. Results from the 128k test should be interesting. Why -1 ? I don't know , maybe its saying something about the quality .

Yes. I remember that hiss or something that exists in the background of some of the lower parts of an album I had. I could tell that it was caused by the encoder, as the original didn't have it and Ogg Vorbis (any encoder) seemed to have it up to q 4. When I pushed it up to 5 (then using the GT3b1 encoder), this noise was gone. I know many say that transparency occurs at q 5 -- 6 but, I think it would depend on what kind of audio's being sent to it, right? What if the particular piece wasn't transparent by, say, q 9? Would that mean that the encoder has a problem or just that the piece is very demanding? Something else I noticed as I dropped the bitrate much lower is that stereo separation becomes haphazard but still everything's impressive for such low (around -1 -- 1) q settings. Finally, by the time I reach q -1, I have problems seeking through the file whilst using Winamp 5.02. The timer is showing a time that doesn't match what's being played. Performance is much better than what I get with MP3, particularly at the lower bitrates. However, I still use MP3 only because my hardware (iRiver iMP-550) doesn't support every Vorbis bitrate (96 -- 256 kbps; any audio outside this range isn't played).

 
SimplePortal 1.0.0 RC1 © 2008-2019