[openal] Programmatic access to sounds played through OpenAL

[Chris Robinson]

On Thu, 3 Sep 2020 19:27:24 -0700
Karkala Hegde <khegde at usc.edu> wrote:

> Hi Chris,
> Thanks for your answer. I followed the same instructions you
> mentioned and tried writing the samples to a wave file, I assume I
> can create the buffer like so:
>     int test_buffer[1024];
> and then populate it using the code you provided.
> 
> You mentioned that if we concatenated all the samples from each frame
> and played at normal speed it would sound the same. How do you
> convert the single dimension buffer array to a 2 channel stereo wav
> file (or two wav files, one for each channel)?

Normally audio is stored with interleaved channels, meaning the samples
for each channel follow consecutively.

buffer[0] | buffer[1] | buffer[2] | buffer[3] | ...
--------------------------------------------------
  Left-0  |  Right-0  |   Left-1  |  Right-1  | ...


This is how the loopback extension gives the audio. So if you have a
16-bit stereo sound stored in a

   short buffer[1024];

then to access sample 'n' of channel 'm', you would do

   buffer[n*2 + m]

where '2' can be replaced with the number of channels if different. You
can alternatively declare a multidimensional array as long as the
channel count is fixed:

   short stereo_buffer[512][2];
   ...
   stereo_buffer[n][m]

but that can only work if the channel count is known at compile time.
It won't work correctly if you get mono or other non-stereo data.

To concatenate interleaved audio buffers, just copy all the bytes
from each buffer as-is one after the other. The last channel's last
sample from the first buffer will be followed by the first channel's
first sample from the second buffer, which is correct.