home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!dtix!oasys!curt
- From: curt@oasys.dt.navy.mil (Curt Welch)
- Newsgroups: rec.audio
- Subject: Re: MD recording quality
- Message-ID: <27742@oasys.dt.navy.mil>
- Date: 20 Nov 92 20:04:58 GMT
- References: <1992Nov20.175144.15222@csqx.cs.rhbnc.ac.uk>
- Reply-To: curt@oasys.dt.navy.mil (Curt Welch)
- Organization: Carderock Division, NSWC, Bethesda, MD
- Lines: 35
-
- In rec.audio, andy@csqx.cs.rhbnc.ac.uk writes:
- >Something that I haven't seen mentioned before in this group is the fact
- >that the quality of MiniDisc home and studio recording will inevitably
- >differ. ...
-
- >The ATRAC converter to be found in consumer
- >products, especially the smaller personal units where the power consumption
- >of the ATRAC chipset is significant, will not be as sophisticated as that
- >found in the professional unit.
-
- This is something else I've heard and find surprising about these new
- data compression algorithms. I can understand how a professional ATRAC
- unit could have better A/D converters, and have extra features like
- 48K to 44.1K sampling rate conversion, and extra digital input formats,
- but I wouldn't expect the ATRAC encoding to be different at all.
-
- I would expect the ATRAC algorithm to be fixed. Either you do it correclty
- or you don't. But this must not be the case. Can anyone explain more about
- how this works and how (and why) the algorithm would be implemented
- differently on different units?
-
- The limited descriptions that I've seen of the algorithm seem to imply that
- the music is broken into short segments, and that an FFT is done on these
- segments. The resulting set of numbers is then scanned, and some of the
- data is thrown away, and the remaining info is then encoded, with varying
- degrees of accuracy. The music is basically stored on the disk in the FFT
- form.
-
- So what changes in the more expensive converters? Is there some type of
- trade off on the cpu time required to analyze the the data when deciding
- what to throw away and what to keep? Or does it have more to do with
- deciding how to encode the frequence data to maximise the information
- saved in the fixed block on the disk?
-
- Curt
-