Bug/aac missing timeline

[dedobbin]

Fixes #196

[pdeljanov]

Could you please test this with the following command: cat /path/to/your/aac/file.aac | symphonia-play -.

I don't think the seek will work if the input is not a seekable file! You will need to use the seek buffer in these cases.

[dedobbin]

Alright. As discussed, it now guarantees being rewindable , getting rid of the nasty panic and making it possible to set a max bytes to sample the approximate frames

[pdeljanov]

Getting very good results with this last set of changes!

[dedobbin]

Alright, fixed all CI problems

[pdeljanov]

This seems to cause errors when testing with symphonia-check. I'm assuming we're dropping a few initial packets for some reason...

[pdeljanov]

This is a bit strange to me. You are attempting to seek to and examine 3 location in the file (guessing because you're doing total_len / 3), but you start from the middle of the file (new_pos = total_len / 2) which means you are skipping half of the file.

Assuming you want to sample at points a, b, and c, shouldn't you calculate step = (total_len - original_pos) / 4, and start at new_pos = orignal_pos + step? Diagram below.

         a        b        c
|        |        |        |        |
^ original_pos                      ^ total_len

[pdeljanov]

This loop could also be made into a for loop:

for new_pos in (original_pos..total_len).step_by(step).skip(1) {

No need for mutable variables!

[pdeljanov]

500 is a lot of frames! That's over 11 seconds of audio at 44.1 kHz. Can we do a magnitude less?

[dedobbin]

Yeah that seems on the extreme side. I was not sure what numbers to plug in, i lowered it now and it still gives acceptable output to me. Let me know if this works for you or how we should go about finding a sweet spot.