[MPlayer-users] Thoughts on "pre-decoding" of compressed sound?

[Richard Bromley]

When I play DVDs with Dolby Digital sound, I nearly always observe
framedrops occurring periodically, typically with moderate to high
frequency, and with short but noticeable intervals. That is to say, I
observe long periods of smooth video (1-2 minutes in length), broken
by short intervals of relatively jerky video (2-3 seconds in length).
It is my experience that the -autosync parameter is ineffective in
regards to this problem, merely offsetting the cycle by delaying the
first appearance of jerkiness, but not reducing the frequency or
interval length of the framedrops by any apparent degree.

Further, I have observed that when playing DVD videos with DTS sound,
said framedrops occur with markedly reduced frequency and/or with a
smaller interval, providing for a vastly improved visual experience.

I have supposed that this is the case because, although the bitrate of
DTS is higher than that of Dolby Digital, the DTS codec is less
CPU-intensive than the Dolby Digital codec because its compression
algorithms, as I understand it, are not as sophisticated.

This has led me to contemplate whether or not it would be worthwhile
to decode a Dolby Digital soundtrack to a multichannel PCM wave file
before playing a DVD, and then to play the movie with the -audiofile
parameter, to instruct MPlayer to use the "pre-decoded", uncompressed
sound data. However, I am uncertain as to the value of this endeavor.

Firstly, it is my uninformed guess that I/O would increase, due to the
massive bitrate of 6 uncompressed, 48 KHz, 16-bit channels of sound.
Because I am not educated in computer science, I am uncertain as to
whether or not, despite said increase in I/O, any CPU power would be
freed, consequently improving smoothness, or if, instead, this
increase in I/O would and of itself cause an increase in CPU
consumption, thus counterbalancing the lack of CPU utilization for
audio decoding purposes.

Secondly, if I have guessed correctly, audio/video synchronization is
not possilbe with an external audio file scheme. Thus, any observed
improvement in smoothness - providing the high I/O rate would allow
for one, as I addressed above - would be confounded by the possibility
that the lack of synchronization had partial, or full, responsibility
for such a positive result. Obviously, a lack of synchronization could
also have negative, and visible, consequences, increasing in severity
with time, and possibly equalling or exceeding in detrimental impact
the original framedrops, themselves.

To summarize: for playing DVDs, is it at all beneficial to do the following?

mplayer -ao pcm:fast:file=sound.wav dvd:// &&
mplayer -audiofile sound.wav dvd://

Thanks for your time. (I apologize for the wordiness of my post, but
this is a subject I enjoy. I am of course open to any sort of
educational information, advice, etc. that can be offered in regards
to this post's subject matter.)