Audio Notes

notes.txt

 0               1               2             3
 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 2 3 4 5 6 7 0 1 2 3 4 5 6 7
+---+-+-+-------+---------------+-----------------------------+
|V  |P|X| CSRC  |M|Payload type |          Sequence           |
+---+-+-+-------+---------------+-----------------------------+
|                        Timestamp                            |
+-------------------------------------------------------------+
|                          SSRC                               |
+-------------------------------------------------------------+
|                       Audio Data ...                        |
+-------------------------------------------------------------+


V (Version):  2 bits

    RTP version.
    Always 2 (10) for Discord.

P (Padding):  1 bit

    Whether padding is used in data.
    Always 0 for Discord.

X (Extensions):  1 bit

    Whether RTP extensions are used.
    Always 0 for Discord.

CSRC:  4 bits

    Number of combined audio streams.
    Discord does not combine audio streams. Always 0.

M (Marker):  1 bit

    Custom use for applications.
    Unused by Discord. Always 0.

Payload type:  7 bits

    Defines the RTP payload profile.
    Discord uses 8 (001000).

Sequence:  2 bytes

    Incremented for each RTP packet sent. Used to detect packet loss.

Timestamp:  4 bytes

    Used by the receiver to play back the received samples at appropriate time and interval.
    Discord expects a sample rate of 48kHz, and a frame size of 20 milliseconds. Therefore, the timestamp is incremented by 960 for every packet sent. 
    (48000Hz) / (1s / 20ms) = 960

SSRC:  4 bytes

    Uniquely identifies the source of a stream.
    Each user is a source so they each have a unique SSRC.

Audio Data:  1920 bytes

    Opus-encoded audio data encrypted according to the set encryption mode.
    960 * 2 channels = 1920


== Notes ==
First byte consiting of V,P,X,CSRC is always 0x80
Second byte consiting of M,PT is always 0x78