This is nonsense! I've been using linux for ten years and there is no such thing as being "ready to use linux on the desktop". With respect to OS X, Linux is a huge step backwards in productivity. Linux as a desktop system is inconsistent, amateurish and butt ugly. The only reason I stuck with Linux so long is that I love the CLI, Emacs and the ability to run my simulations on 50 machines simultaneously due to the multi-user paradigm. However, creating a scientific poster or a presentation was always an ordeal as the tools available are ancient relics.
After I got my PB 1Ghz I could never, ever work on a linux box again. It just has to be tried, and then it is no way back. I still buy PCs with linux and stick in the computerroom -- no monitor, mouse or keyboard. Just lots of memory and and fast cpu's that can run the CLI programs I develop under OS X.
I guess many will look at this posting as pure trolling, but I really mean it - many linux users really are in denial. Linux is great for many things, but as a desktop system it is a useless, sorry, unproductive mess... I'd still use Linux instead of windows, as most of what I do is better suited to this platform, but with OS X I can do all that, and I have a great desktop OS as well. Unbeatable...
Speech recognition is extremely CPU and memory intensive by its very nature. Depending on the vocabulary size and the complexity of the language model used (a language model is a statistical model that gives you the probability of a word given the previous N words in the sentence), most research code uses 200-400 Mb of memory and needs really fast CPUs to achieve real time recognition. Although I figure MS has a bag of heuristics up their sleeves theres only so much trickery you can do before the error rate becomes useless.
Anyway, ASR is pretty useful for some people and as long as nothing else needs your RAM and CPU I dont think MS' system should be much worse in terms of resource demands than e.g. IBMs ViaVoice.
However, turning on speech recognition by default really tells you that the people selling these preinstalled machines really don't know what they're doing.
Slashdot Mirror
This is nonsense! I've been using linux for ten years and there is no such thing as being "ready to use linux on the desktop". With respect to OS X, Linux is a huge step backwards in productivity. Linux as a desktop system is inconsistent, amateurish and butt ugly. The only reason I stuck with Linux so long is that I love the CLI, Emacs and the ability to run my simulations on 50 machines simultaneously due to the multi-user paradigm. However, creating a scientific poster or a presentation was always an ordeal as the tools available are ancient relics.
After I got my PB 1Ghz I could never, ever work on a linux box again. It just has to be tried, and then it is no way back. I still buy PCs with linux and stick in the computerroom -- no monitor, mouse or keyboard. Just lots of memory and and fast cpu's that can run the CLI programs I develop under OS X.
I guess many will look at this posting as pure trolling, but I really mean it - many linux users really are in denial. Linux is great for many things, but as a desktop system it is a useless, sorry, unproductive mess... I'd still use Linux instead of windows, as most of what I do is better suited to this platform, but with OS X I can do all that, and I have a great desktop OS as well. Unbeatable...
Speech recognition is extremely CPU and memory intensive by its very nature. Depending on the vocabulary size and the complexity of the language model used (a language model is a statistical model that gives you the probability of a word given the previous N words in the sentence), most research code uses 200-400 Mb of memory and needs really fast CPUs to achieve real time recognition. Although I figure MS has a bag of heuristics up their sleeves theres only so much trickery you can do before the error rate becomes useless.
Anyway, ASR is pretty useful for some people and as long as nothing else needs your RAM and CPU I dont think MS' system should be much worse in terms of resource demands than e.g. IBMs ViaVoice.
However, turning on speech recognition by default really tells you that the people selling these preinstalled machines really don't know what they're doing.