I'm sorry if this is just bullshit but I'd like to put forward a thought I've had...

Would it be useful or even make sense for a switch to be implemented to vorbis where, for example if a sound sample (or part of a sample) can be encoded losslessly without going over a certain bitrate (say, 500kbps) it will encoded it with lossless filters... if lossless needs more than 500kbps then, it encodes it ''quasi-lossless'' where it is totally transparent and also is optimal for transcoding...?

Since FLAC manages to compress most files down to ~60% (847kbps) it might not be worth the bother but perhaps its something to develop on?

Vorbis is not designed for lossless encoding. If you wanted to create a codec like that it'd be better to start from FLAC or another lossless codec, though I can't imagine what you would use it for.

Wavpack has a lossy mode that sounds kinda like what you want, except there's no automatic process to switch between lossy or lossless based on user set bitrate thresholds.

You could do it manually with foobar2k: compress the files in lossless mode for example, then just sort by bitrate and recompress the ones with the highest bitrates in lossy mode.

I'd concur with the poster above - use wavpack hybrid lossy mode if you want something high-bitrate and "near-lossless" that doesn't suffer from typical psychoacoustic codec shortcomings. All wavpack hybrid does is add quantization noise to the signal. Unlike MP3, AAC or Vorbis, you won't get transient problems and pre-echo, so if the added white noise is masked or below your threshold of hearing, the resulting signal will be completely transparent to the original. 320-384kbps seems to work very well, which is even less than your 500kbps target.

Of course, transparency is achievable at much lower bitrates with the psychoacoustic codecs. Have you tried Vorbis -q6 or LAME -V2?

Yea, vorbis -q6 is pretty mych transparent to me, but I just had funny notions of whether its future proof in that some time in the future if my kids wanted to listen to my lossy collection they wouldn't be complaining about sound quality... erm.. never mind...

Well, use lossless then
Why do half a perfect job?

I would worry more about backwards compatibility if Vorbis 2.0 ever get's developed and reaches the light of day. Open source projects work like this. They will just be around forever until somebody decides to resurrect them 5 years down the road and implement new things into them. If you stick with lossless you will never have anything to worry about.

I'd say lossless formats are more future secure than lossy. With Wavpack and Optimfrog you can have 'hybrid' encodings for lossy + full lossless restoration. Even without the correction files you have fully featured formats with proper seeking, high-res , multichannel and free of psychoacoustic artifacts. It is much harder for wavpack and optimfrog lossy to die because they are also lossless formats. Even in the worse case scenario, you can transcode the 500k lossy file into a LOSSLESS format and get practicaly the same filesize as going from WAV > lossless. That may be overkill , but it is a very high level of security and virtually futureproof.

I wouldn't worry. People won't evolve ultrasonic hearing any time soon. People's perception is likely to stay the same, so perceptual coding (mostly synonymous with lossy compression) should be safe for the time being.

I agree. The really unlikely thing though, is that kids will want to listen to their parent's music in the first place

Wouldn't encoding at a VBR (variable bit rate) do what you're asking? You can specify the quality level you want. If I understand this orrectly if you set it to VBR quality level 7 everything that can be will be encoded at lower bit rates but it will never go above level 7.