dpp::discord_voice_client Class Reference

Implements a discord voice connection. Each discord_voice_client connects to one voice channel and derives from a websocket client. More...

#include <discordvoiceclient.h>

Inheritance diagram for dpp::discord_voice_client:

Collaboration diagram for dpp::discord_voice_client:

Public Types
enum	send_audio_type_t { satype_recorded_audio , satype_live_audio , satype_overlap_audio }
	The audio type to be sent. The default type is recorded audio. More...

Public Member Functions
void	set_user_gain (snowflake user_id, float factor)
	Sets the gain for the specified user. More...
virtual void	log (dpp::loglevel severity, const std::string &msg) const
	Log a message to whatever log the user is using. The logged message is passed up the chain to the on_log event in user code which can then do whatever it wants to do with it. More...
virtual void	one_second_timer ()
	Fires every second from the underlying socket I/O loop, used for sending heartbeats. More...
bool	is_ready ()
	voice client is ready to stream audio. The voice client is considered ready if it has a secret key. More...
bool	is_connected ()
	Returns true if the voice client is connected to the websocket. More...
dpp::utility::uptime	get_uptime ()
	Returns the connection time of the voice client. More...
	discord_voice_client (dpp::cluster *_cluster, snowflake _channel_id, snowflake _server_id, const std::string &_token, const std::string &_session_id, const std::string &_host)
virtual	~discord_voice_client ()
	Destroy the discord voice client object. More...
virtual bool	handle_frame (const std::string &buffer)
	Handle JSON from the websocket. More...
virtual void	error (uint32_t errorcode)
	Handle a websocket error. More...
void	run ()
	Start and monitor I/O loop. More...
discord_voice_client &	send_audio_raw (uint16_t *audio_data, const size_t length)
	Send raw audio to the voice channel. More...
discord_voice_client &	send_audio_opus (uint8_t *opus_packet, const size_t length, uint64_t duration)
	Send opus packets to the voice channel. More...
discord_voice_client &	send_audio_opus (uint8_t *opus_packet, const size_t length)
	Send opus packets to the voice channel. More...
discord_voice_client &	send_silence (const uint64_t duration)
	Send silence to the voice channel. More...
discord_voice_client &	set_send_audio_type (send_audio_type_t type)
	Sets the audio type that will be sent with send_audio_* methods. More...
discord_voice_client &	set_timescale (uint64_t new_timescale)
	Set the timescale in nanoseconds. More...
uint64_t	get_timescale ()
	Get the current timescale, this will default to 1000000 which means 1 millisecond. More...
discord_voice_client &	speak ()
	Mark the voice connection as 'speaking'. This sends a JSON message to the voice websocket which tells discord that the user is speaking. The library automatically calls this for you whenever you send audio. More...
discord_voice_client &	pause_audio (bool pause)
	Pause sending of audio. More...
discord_voice_client &	stop_audio ()
	Immediately stop all audio. Clears the packet queue. More...
bool	is_playing ()
	Returns true if we are playing audio. More...
float	get_secs_remaining ()
	Get the number of seconds remaining of the audio output buffer. More...
uint32_t	get_tracks_remaining ()
	Get the number of tracks remaining in the output buffer. This is calculated by the number of track markers plus one. More...
dpp::utility::uptime	get_remaining ()
	Get the time remaining to send the audio output buffer in hours:minutes:seconds. More...
discord_voice_client &	insert_marker (const std::string &metadata="")
	Insert a track marker into the audio output buffer. A track marker is an arbitrary flag in the buffer contents that indicates the end of some block of audio of significance to the sender. This may be a song from a streaming site, or some voice audio/speech, a sound effect, or whatever you choose. You can later skip to the next marker using the dpp::discord_voice_client::skip_to_next_marker function. More...
discord_voice_client &	skip_to_next_marker ()
	Skip tp the next track marker, previously inserted by using the dpp::discord_voice_client::insert_marker function. If there are no markers in the output buffer, then this skips to the end of the buffer and is equivalent to the dpp::discord_voice_client::stop_audio function. More...
const std::vector< std::string >	get_marker_metadata ()
	Get the metadata string associated with each inserted marker. More...
bool	is_paused ()
	Returns true if the audio is paused. You can unpause with dpp::discord_voice_client::pause_audio. More...
std::string	discover_ip ()
	Discord external IP detection. More...
virtual void	write (const std::string &data)
	Write to websocket. Encapsulates data in frames if the status is CONNECTED. More...
virtual bool	handle_buffer (std::string &buffer)
	Processes incoming frames from the SSL socket input buffer. More...
virtual void	close ()
	Close websocket. More...
void	send_close_packet ()
	Send OP_CLOSE error code 1000 to the other side of the connection. This indicates graceful close. More...
uint64_t	get_bytes_out ()
	Get the bytes out objectGet total bytes sent. More...
uint64_t	get_bytes_in ()
	Get total bytes received. More...
std::string	get_cipher ()
	Get SSL cipher name. More...
void	read_loop ()
	Nonblocking I/O loop. More...

Public Attributes
class dpp::cluster *	creator
	Owning cluster. More...
bool	terminating
	True when the thread is shutting down. More...
float	end_gain
	The gain value for the end of the current voice iteration. More...
float	current_gain
	The gain value for the current voice iteration. More...
float	increment
	The amount to increment each successive sample for, for the current voice iteration. More...
uint32_t	heartbeat_interval
	Heartbeat interval for sending heartbeat keepalive. More...
time_t	last_heartbeat
	Last voice channel websocket heartbeat. More...
std::thread::native_handle_type	thread_id
	Thread ID. More...
std::string	token
	Discord voice session token. More...
std::string	sessionid
	Discord voice session id. More...
snowflake	server_id
	Server ID. More...
moving_averager	moving_average
	Moving averager. More...
snowflake	channel_id
	Channel ID. More...
enum dpp::discord_voice_client::send_audio_type_t	send_audio_type = satype_recorded_audio
socket_callback_t	custom_readable_fd
	Attaching an additional file descriptor to this function will send notifications when there is data to read. More...
socket_callback_t	custom_writeable_fd
	Attaching an additional file descriptor to this function will send notifications when you are able to write to the socket. More...
socket_notification_t	custom_readable_ready
	This event will be called when you can read from the custom fd. More...
socket_notification_t	custom_writeable_ready
	This event will be called when you can write to a custom fd. More...
bool	keepalive
	True if we are keeping the connection alive after it has finished. More...

Static Public Attributes
static bool	sodium_initialised
	This needs to be static, we only initialise libsodium once per program start, so initialising it on first use in a voice connection is best. More...

Protected Member Functions
virtual void	connect ()
	(Re)connect More...
ws_state	get_state ()
	Get websocket state. More...

Protected Attributes
std::string	buffer
	Input buffer received from socket. More...
std::string	obuffer
	Output buffer for sending to socket. More...
bool	nonblocking
	True if in nonblocking mode. The socket switches to nonblocking mode once ReadLoop is called. More...
dpp::socket	sfd
	Raw file descriptor of connection. More...
openssl_connection *	ssl
	Openssl opaque contexts. More...
std::string	cipher
	SSL cipher in use. More...
time_t	last_tick
	For timers. More...
std::string	hostname
	Hostname connected to. More...
uint64_t	bytes_out
	Bytes out. More...
uint64_t	bytes_in
	Bytes in. More...
bool	plaintext
	True for a plain text connection. More...
bool	make_new
	True if we are establishing a new connection, false if otherwise. More...

Detailed Description

Implements a discord voice connection. Each discord_voice_client connects to one voice channel and derives from a websocket client.

Member Enumeration Documentation

send_audio_type_t

enum dpp::discord_voice_client::send_audio_type_t

The audio type to be sent. The default type is recorded audio.

If the audio is recorded, the sending of audio packets is throttled. Otherwise, if the audio is live, the sending is not throttled.

Discord voice engine is expecting audio data as if they were from some audio device, e.g. microphone, where the data become available as they get captured from the audio device.

In case of recorded audio, unlike from a device, the audio data are usually instantly available in large chunks. Throttling is needed to simulate audio data coming from an audio device. In case of live audio, the throttling is by nature, so no extra throttling is needed.

Using live audio mode for recorded audio can cause Discord to skip audio data because Discord does not expect to receive, say, 3 minutes' worth of audio data in 1 second.

There are some inaccuracies in the throttling method used by the recorded audio mode on some systems (mainly Windows) which causes gaps and stutters in the resulting audio stream. The overlap audio mode provides a different implementation that fixes the issue. This method is slightly more CPU intensive, and should only be used if you encounter issues with recorded audio on your system.

Use discord_voice_client::set_send_audio_type to change this value as it ensures thread safety.

Enumerator
satype_recorded_audio
satype_live_audio
satype_overlap_audio

Constructor & Destructor Documentation

discord_voice_client()

dpp::discord_voice_client::discord_voice_client	(	dpp::cluster *	_cluster,
		snowflake	_channel_id,
		snowflake	_server_id,
		const std::string &	_token,
		const std::string &	_session_id,
		const std::string &	_host
	)

Constructor takes shard id, max shards and token.

Parameters

_cluster	The cluster which owns this voice connection, for related logging, REST requests etc
_channel_id	The channel id to identify the voice connection as
_server_id	The server id (guild id) to identify the voice connection as
_token	The voice session token to use for identifying to the websocket
_session_id	The voice session id to identify with
_host	The voice server hostname to connect to (hostname:port format)

Exceptions

dpp::voice_exception

Sodium or Opus failed to initialise, or D++ is not compiled with voice support

~discord_voice_client()

virtual dpp::discord_voice_client::~discord_voice_client ( )

virtual

Destroy the discord voice client object.

Member Function Documentation

close()

virtual void dpp::websocket_client::close ( )

virtualinherited

Close websocket.

Reimplemented from dpp::ssl_client.

connect()

virtual void dpp::websocket_client::connect ( )

protectedvirtualinherited

(Re)connect

Reimplemented from dpp::ssl_client.

discover_ip()

std::string dpp::discord_voice_client::discover_ip

(

)

Discord external IP detection.

Returns

std::string Your external IP address

Note

This is a blocking operation that waits for a single packet from Discord's voice servers.

error()

virtual void dpp::discord_voice_client::error ( uint32_t  errorcode )

virtual

Handle a websocket error.

Parameters

errorcode

The error returned from the websocket

Reimplemented from dpp::websocket_client.

get_bytes_in()

uint64_t dpp::ssl_client::get_bytes_in ( )

inherited

Get total bytes received.

Returns

uint64_t bytes received

get_bytes_out()

uint64_t dpp::ssl_client::get_bytes_out ( )

inherited

Get the bytes out objectGet total bytes sent.

Returns

uint64_t bytes sent

get_cipher()

std::string dpp::ssl_client::get_cipher ( )

inherited

Get SSL cipher name.

Returns

std::string ssl cipher name

get_marker_metadata()

const std::vector< std::string > dpp::discord_voice_client::get_marker_metadata

(

)

Get the metadata string associated with each inserted marker.

Returns

const std::vector<std::string>& list of metadata strings

get_remaining()

dpp::utility::uptime dpp::discord_voice_client::get_remaining

(

)

Get the time remaining to send the audio output buffer in hours:minutes:seconds.

Returns

dpp::utility::uptime length of buffer

get_secs_remaining()

float dpp::discord_voice_client::get_secs_remaining

(

)

Get the number of seconds remaining of the audio output buffer.

Returns

float number of seconds remaining

get_state()

ws_state dpp::websocket_client::get_state ( )

protectedinherited

Get websocket state.

Returns

websocket state

get_timescale()

uint64_t dpp::discord_voice_client::get_timescale

(

)

Get the current timescale, this will default to 1000000 which means 1 millisecond.

Returns

uint64_t timescale in nanoseconds

get_tracks_remaining()

uint32_t dpp::discord_voice_client::get_tracks_remaining

(

)

Get the number of tracks remaining in the output buffer. This is calculated by the number of track markers plus one.

Returns

uint32_t Number of tracks in the buffer

get_uptime()

dpp::utility::uptime dpp::discord_voice_client::get_uptime

(

)

Returns the connection time of the voice client.

Returns

dpp::utility::uptime Detail of how long the voice client has been connected for

handle_buffer()

virtual bool dpp::websocket_client::handle_buffer ( std::string &  buffer )

virtualinherited

Processes incoming frames from the SSL socket input buffer.

Parameters

buffer

The buffer contents. Can modify this value removing the head elements when processed.

Reimplemented from dpp::ssl_client.

handle_frame()

virtual bool dpp::discord_voice_client::handle_frame ( const std::string &  buffer )

virtual

Handle JSON from the websocket.

Parameters

buffer

The entire buffer content from the websocket client

Returns

bool True if a frame has been handled

Exceptions

dpp::exception

If there was an error processing the frame, or connection to UDP socket failed

Reimplemented from dpp::websocket_client.

insert_marker()

discord_voice_client & dpp::discord_voice_client::insert_marker

(

const std::string &

metadata = ""

)

Insert a track marker into the audio output buffer. A track marker is an arbitrary flag in the buffer contents that indicates the end of some block of audio of significance to the sender. This may be a song from a streaming site, or some voice audio/speech, a sound effect, or whatever you choose. You can later skip to the next marker using the dpp::discord_voice_client::skip_to_next_marker function.

Parameters

metadata

Arbitrary information related to this track

Returns

reference to self

is_connected()

bool dpp::discord_voice_client::is_connected

(

)

Returns true if the voice client is connected to the websocket.

Returns

True if connected

is_paused()

bool dpp::discord_voice_client::is_paused

(

)

Returns true if the audio is paused. You can unpause with dpp::discord_voice_client::pause_audio.

Returns

true if paused

is_playing()

bool dpp::discord_voice_client::is_playing

(

)

Returns true if we are playing audio.

Returns

true if audio is playing

is_ready()

bool dpp::discord_voice_client::is_ready

(

)

voice client is ready to stream audio. The voice client is considered ready if it has a secret key.

Returns

true if ready to stream audio

log()

virtual void dpp::discord_voice_client::log ( dpp::loglevel  severity, const std::string &  msg  ) const

virtual

Log a message to whatever log the user is using. The logged message is passed up the chain to the on_log event in user code which can then do whatever it wants to do with it.

Parameters

severity	The log level from dpp::loglevel
msg	The log message to output

Reimplemented from dpp::ssl_client.

one_second_timer()

virtual void dpp::discord_voice_client::one_second_timer ( )

virtual

Fires every second from the underlying socket I/O loop, used for sending heartbeats.

Exceptions

dpp::exception

if the socket needs to disconnect

Reimplemented from dpp::websocket_client.

pause_audio()

discord_voice_client & dpp::discord_voice_client::pause_audio

(

bool

pause

)

Pause sending of audio.

Parameters

pause

True to pause, false to resume

Returns

reference to self

read_loop()

void dpp::ssl_client::read_loop ( )

inherited

Nonblocking I/O loop.

Exceptions

std::exception

Any std::exception (or derivative) thrown from read_loop() causes reconnection of the shard

run()

void dpp::discord_voice_client::run

(

)

Start and monitor I/O loop.

send_audio_opus() [1/2]

discord_voice_client & dpp::discord_voice_client::send_audio_opus	(	uint8_t *	opus_packet,
		const size_t	length
	)

Send opus packets to the voice channel.

Some containers such as .ogg may contain OPUS encoded data already. In this case, we don't need to encode the frames using opus here. We can bypass the codec, only applying libsodium to the stream.

Duration is calculated internally

Parameters

opus_packet	Opus packets. Discord expects opus frames to be encoded at 48000Hz
length	The length of the audio data.

Returns

discord_voice_client& Reference to self

Note

It is your responsibility to ensure that packets of data sent to send_audio are correctly repacketized for streaming, e.g. that audio frames are not too large or contain an incorrect format. Discord will still expect the same frequency and bit width of audio and the same signedness.

Exceptions

dpp::voice_exception

If data length is invalid or voice support not compiled into D++

send_audio_opus() [2/2]

discord_voice_client & dpp::discord_voice_client::send_audio_opus	(	uint8_t *	opus_packet,
		const size_t	length,
		uint64_t	duration
	)

Send opus packets to the voice channel.

Some containers such as .ogg may contain OPUS encoded data already. In this case, we don't need to encode the frames using opus here. We can bypass the codec, only applying libsodium to the stream.

Parameters

opus_packet	Opus packets. Discord expects opus frames to be encoded at 48000Hz
length	The length of the audio data.
duration	Generally duration is 2.5, 5, 10, 20, 40 or 60 if the timescale is 1000000 (1ms)

Returns

discord_voice_client& Reference to self

Note

It is your responsibility to ensure that packets of data sent to send_audio are correctly repacketized for streaming, e.g. that audio frames are not too large or contain an incorrect format. Discord will still expect the same frequency and bit width of audio and the same signedness.

Exceptions

dpp::voice_exception

If data length is invalid or voice support not compiled into D++

send_audio_raw()

discord_voice_client & dpp::discord_voice_client::send_audio_raw	(	uint16_t *	audio_data,
		const size_t	length
	)

Send raw audio to the voice channel.

You should send an audio packet of 11520 bytes. Note that this function can be costly as it has to opus encode the PCM audio on the fly, and also encrypt it with libsodium.

Note

Because this function encrypts and encodes packets before pushing them onto the output queue, if you have a complete stream ready to send and know its length it is advisable to call this method multiple times to enqueue the entire stream audio so that it is all encoded at once (unless you have set use_opus to false). Constantly calling this from the dpp::on_voice_buffer_send callback can and will eat a TON of cpu!

Parameters

audio_data

Raw PCM audio data. Channels are interleaved, with each channel's amplitude being a 16 bit value.

The audio data should be 48000Hz signed 16 bit audio.

Parameters

length

The length of the audio data. The length should be a multiple of 4 (2x 16 bit stereo channels) with a maximum length of 11520, which is a complete opus frame at highest quality.

Returns

discord_voice_client& Reference to self

Exceptions

dpp::voice_exception

If data length is invalid or voice support not compiled into D++

send_close_packet()

void dpp::websocket_client::send_close_packet ( )

inherited

Send OP_CLOSE error code 1000 to the other side of the connection. This indicates graceful close.

send_silence()

discord_voice_client & dpp::discord_voice_client::send_silence

(

const uint64_t

duration

)

Send silence to the voice channel.

Parameters

duration

How long to send silence for. With the standard timescale this is in milliseconds. Allowed values are 2.5, 5, 10, 20, 40 or 60 milliseconds.

Returns

discord_voice_client& Reference to self

Exceptions

dpp::voice_exception

if voice support is not compiled into D++

set_send_audio_type()

discord_voice_client & dpp::discord_voice_client::set_send_audio_type

(

send_audio_type_t

type

)

Sets the audio type that will be sent with send_audio_* methods.

See also

send_audio_type_t

set_timescale()

discord_voice_client & dpp::discord_voice_client::set_timescale

(

uint64_t

new_timescale

)

Set the timescale in nanoseconds.

Parameters

new_timescale

Timescale to set. This defaults to 1000000, which means 1 millisecond.

Returns

discord_voice_client& Reference to self

Exceptions

dpp::voice_exception

If data length is invalid or voice support not compiled into D++

set_user_gain()

void dpp::discord_voice_client::set_user_gain	(	snowflake	user_id,
		float	factor
	)

Sets the gain for the specified user.

Similar to the User Volume slider, controls the listening volume per user. Uses native Opus gain control, so clients don't have to perform extra audio processing.

The gain setting will affect the both individual and combined voice audio.

The gain value can also be set even before the user connects to the voice channel.

Parameters

user_id	The ID of the user where the gain is to be controlled.
factor	Nonnegative factor to scale the amplitude by, where 1.f reverts to the default volume.

skip_to_next_marker()

discord_voice_client & dpp::discord_voice_client::skip_to_next_marker

(

)

Skip tp the next track marker, previously inserted by using the dpp::discord_voice_client::insert_marker function. If there are no markers in the output buffer, then this skips to the end of the buffer and is equivalent to the dpp::discord_voice_client::stop_audio function.

Note

It is possible to use this function while the output stream is paused.

Returns

reference to self

speak()

discord_voice_client & dpp::discord_voice_client::speak

(

)

Mark the voice connection as 'speaking'. This sends a JSON message to the voice websocket which tells discord that the user is speaking. The library automatically calls this for you whenever you send audio.

Returns

discord_voice_client& Reference to self

stop_audio()

discord_voice_client & dpp::discord_voice_client::stop_audio

(

)

Immediately stop all audio. Clears the packet queue.

Returns

reference to self

write()

virtual void dpp::websocket_client::write ( const std::string &  data )

virtualinherited

Write to websocket. Encapsulates data in frames if the status is CONNECTED.

Parameters

data	The data to send.

Reimplemented from dpp::ssl_client.

Member Data Documentation

buffer

std::string dpp::ssl_client::buffer

protectedinherited

Input buffer received from socket.

bytes_in

uint64_t dpp::ssl_client::bytes_in

protectedinherited

Bytes in.

bytes_out

uint64_t dpp::ssl_client::bytes_out

protectedinherited

Bytes out.

channel_id

snowflake dpp::discord_voice_client::channel_id

Channel ID.

cipher

std::string dpp::ssl_client::cipher

protectedinherited

SSL cipher in use.

creator

class dpp::cluster* dpp::discord_voice_client::creator

Owning cluster.

current_gain

float dpp::discord_voice_client::current_gain

The gain value for the current voice iteration.

custom_readable_fd

socket_callback_t dpp::ssl_client::custom_readable_fd

inherited

Attaching an additional file descriptor to this function will send notifications when there is data to read.

NOTE: Only hook this if you NEED it as it can increase CPU usage of the thread! Returning -1 means that you don't want to be notified.

custom_readable_ready

socket_notification_t dpp::ssl_client::custom_readable_ready

inherited

This event will be called when you can read from the custom fd.

custom_writeable_fd

socket_callback_t dpp::ssl_client::custom_writeable_fd

inherited

Attaching an additional file descriptor to this function will send notifications when you are able to write to the socket.

NOTE: Only hook this if you NEED it as it can increase CPU usage of the thread! You should toggle this to -1 when you do not have anything to write otherwise it'll keep triggering repeatedly (it is level triggered).

custom_writeable_ready

socket_notification_t dpp::ssl_client::custom_writeable_ready

inherited

This event will be called when you can write to a custom fd.

end_gain

float dpp::discord_voice_client::end_gain

The gain value for the end of the current voice iteration.

heartbeat_interval

uint32_t dpp::discord_voice_client::heartbeat_interval

Heartbeat interval for sending heartbeat keepalive.

hostname

std::string dpp::ssl_client::hostname

protectedinherited

Hostname connected to.

increment

float dpp::discord_voice_client::increment

The amount to increment each successive sample for, for the current voice iteration.

keepalive

bool dpp::ssl_client::keepalive

inherited

True if we are keeping the connection alive after it has finished.

last_heartbeat

time_t dpp::discord_voice_client::last_heartbeat

Last voice channel websocket heartbeat.

last_tick

time_t dpp::ssl_client::last_tick

protectedinherited

For timers.

make_new

bool dpp::ssl_client::make_new

protectedinherited

True if we are establishing a new connection, false if otherwise.

moving_average

moving_averager dpp::discord_voice_client::moving_average

Moving averager.

nonblocking

bool dpp::ssl_client::nonblocking

protectedinherited

True if in nonblocking mode. The socket switches to nonblocking mode once ReadLoop is called.

obuffer

std::string dpp::ssl_client::obuffer

protectedinherited

Output buffer for sending to socket.

plaintext

bool dpp::ssl_client::plaintext

protectedinherited

True for a plain text connection.

send_audio_type

enum dpp::discord_voice_client::send_audio_type_t dpp::discord_voice_client::send_audio_type = satype_recorded_audio

server_id

snowflake dpp::discord_voice_client::server_id

Server ID.

sessionid

std::string dpp::discord_voice_client::sessionid

Discord voice session id.

sfd

dpp::socket dpp::ssl_client::sfd

protectedinherited

Raw file descriptor of connection.

sodium_initialised

bool dpp::discord_voice_client::sodium_initialised

static

This needs to be static, we only initialise libsodium once per program start, so initialising it on first use in a voice connection is best.

ssl

openssl_connection* dpp::ssl_client::ssl

protectedinherited

Openssl opaque contexts.

terminating

bool dpp::discord_voice_client::terminating

True when the thread is shutting down.

thread_id

std::thread::native_handle_type dpp::discord_voice_client::thread_id

Thread ID.

token

std::string dpp::discord_voice_client::token

Discord voice session token.