Got a few more responses from the Rio engineers. Note, none of these figures have been tested. They are doing the tests right now, and these are just guesses copied from his e-mail to me:
There are two factors that determine battery life, in general:
- Decode CPU consumption
- Bitrate (determines spinup frequency)
In general, the decode consumption is far by the biggest factor in battery
life. I know that FLAC is currently taking more power than MP3, although
when optimised it shouldn't (it's on the to-do list). My power spreadsheet
suggests that if it were the same CPU load as MP3, an 800kbps FLAC would
have about 12 hours battery life, but this hasn't been tested.
OGG will suck a lot more battery life than FLAC, despite the lower bitrates.
I'm estimating a worst-case (for a 200kbps FLAC) as being in the 10 hour
range. Again, this hasn't been tested
The Karma doesn't suffer like the iPod does with large files (or did: are
you sure it still has a bug with big files?). We don't cache on a file basis
as that would just be silly - we cache chunks of playable data, hence there
is no difference between caching 10x1mb files or one 10mb file.
Note (to everyone) that on the first play, tracks will take marginally more
power as the track profiler will be profiling the audio data - subsequent
plays are marginally more power efficient.