[Table of Contents]


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ARSCLIST] Audio History In a Nutshell?



Howard Friedman wrote:
Mike,

MEA CULPA! I assumed KB, but that makes it even 1,000 times more difficult to stuff 6MB or more of information into a CDA file! How do they point? How can one produce a CDA file from an MP3 file, for example? If I put an audio CD with 21 tracks into my drive, my file program tells me that there are 924 bytes on the CD. Doesn't really make much sense, does it, unless you're an audio engineer!

I was prepared to be even more snippy in my reply - but I held off.


The limitations of CD-DA are largely dictated by the pointers and their cousins. For instance, the size of a block is defined (and inviolate) because the pointer points to blocks, each of which holds one seventy-fifth of a second of audio. There are enough bits in the pointer to define the starting block and (I believe) the length of the track in blocks. Would you like to end a track somewhere other than a block boundary? You cannot because the next track begins at the boundary; the pointer does not point into the block but only to its start.

One more time - the last, I hope - one cannot produce a CDA file because there is no CDA file. An audio mastering program decompresses an MP3 to approximately WAV format, converting sample rate and bit depth as necessary to make a bit stream of audio data.* It then fills in any bytes left to the next block boundary (don't ask) and writes the appropriate number of blocks to the disc. In TAO, it goes back and updates the directory to reflect the track it has written. In DAO, it precomputes the directory entries and writes them all before writing the audio sequentially. SAO is much like DAO - but please don't ask.

The directory then has an entry for each track - essentially, its starting point and duration. The OS translates those bytes into a "CDA" - a hex representation which is made to appear as though it was a file.

-----------
* Footnote: The conversions take CPU power and the software is rated to work on some minimum CPU. Therefore, the decompression and any conversions needed are simplified to fit that minimum processor (even though your actual CPU may be much more powerful). So you can record 44.1 ksps even though your original file is 24 ksps - but the fit will be relatively crude. For that reason and others, I always recommend feeding the mastering program what it will feed the CD-DA: redbook audio. If you see how long it takes a good program to resample (even just a factor of two), you will appreciate that the mastering program cannot take the time to do as well - especially if it is trying to write at 16x real time.


Mike
--
mrichter@xxxxxxx
http://www.mrichter.com/


[Subject index] [Index for current month] [Table of Contents]