jackclient

Client that dumps samples directly to the jack client with the jack package.

Data:

SERVER

After initializing, JackClient will register itself with this variable.

FS

Sampling rate of the active server

BLOCKSIZE

Blocksize, or the amount of samples processed by jack per each JackClient.process() call.

QUEUE

Queue to be loaded with frames of BLOCKSIZE audio.

Q_LOCK

Lock that enforces a single writer to the QUEUE at a time.

CONTINUOUS

Event that (when set) signals the sound server should play some sound continuously rather than remain silent by default (eg.

CONTINUOUS_QUEUE

Queue that

CONTINUOUS_LOOP

Event flag that is set when frames dropped into the CONTINUOUS_QUEUE should be looped (eg.

Classes:

JackClient([name])

Client that dumps frames of audio directly into a running jackd client.

SERVER = None

After initializing, JackClient will register itself with this variable.

Type

JackClient

FS = 192000

Sampling rate of the active server

Type

int

BLOCKSIZE = 1024

Blocksize, or the amount of samples processed by jack per each JackClient.process() call.

Type

int

QUEUE = None

Queue to be loaded with frames of BLOCKSIZE audio.

Type

multiprocessing.Queue

PLAY = <multiprocessing.synchronize.Event object at 0x7fc8ecf005d0>

Event used to trigger loading samples from QUEUE, ie. playing.

Type

multiprocessing.Event

STOP = <multiprocessing.synchronize.Event object at 0x7fc899aae910>

Event that is triggered on the end of buffered audio.

Note

NOT an event used to stop audio.

Type

multiprocessing.Event

Q_LOCK = None

Lock that enforces a single writer to the QUEUE at a time.

Type

multiprocessing.Lock

CONTINUOUS = None

Event that (when set) signals the sound server should play some sound continuously rather than remain silent by default (eg. play a background sound).

Type

multiprocessing.Event

CONTINUOUS_QUEUE = None

Queue that

Type

multiprocessing.Queue

CONTINUOUS_LOOP = None

Event flag that is set when frames dropped into the CONTINUOUS_QUEUE should be looped (eg. in the case of stationary background noise), otherwise they are played and then discarded (ie. the sound is continuously generating and submitting samples)

Type

multiprocessing.Event

class JackClient(name='jack_client')[source]

Bases: multiprocessing.context.Process

Client that dumps frames of audio directly into a running jackd client.

When first initialized, sets module level variables above.

Variables
  • name (str) – name of client, default “jack_client”

  • q (Queue) – Queue that stores buffered frames of audio

  • q_lock (Lock) – Lock that manages access to the Queue

  • play_evt (multiprocessing.Event) – Event used to trigger loading samples from QUEUE, ie. playing.

  • stop_evt (multiprocessing.Event) – Event that is triggered on the end of buffered audio.

  • quit_evt (multiprocessing.Event) – Event that causes the process to be terminated.

  • client (jack.Client) – Client to interface with jackd

  • blocksize (int) – The blocksize - ie. samples processed per JackClient.process() call.

  • fs (int) – Sampling rate of client

  • zero_arr (numpy.ndarray) – cached array of zeroes used to fill jackd pipe when not processing audio.

  • continuous_cycle (itertools.cycle) – cycle of frames used for continuous sounds

  • mono_output (bool) – True or False depending on if the number of output channels is 1 or >1, respectively. detected and set in JackClient.boot_server() , initialized to True (which is hopefully harmless)

Parameters

name

Methods:

boot_server()

Called by JackClient.run() to boot the server upon starting the process.

run()

Start the process, boot the server, start processing frames and wait for the end.

quit()

Set the JackClient.quit_evt

process(frames)

Process a frame of audio.

pad(data)

Pad a sound that is not a full chunk length long, either by filling with silence or with a continuous sound, if any.

write_to_outports(data)

Write the sound in data to the outport(s).

boot_server()[source]

Called by JackClient.run() to boot the server upon starting the process.

Activates the client and connects it to the physical speaker outputs as determined by prefs.get(‘OUTCHANNELS’).

This is the interpretation of OUTCHANNELS: * empty string

‘mono’ audio: the same sound is always played to all channels. Connect a single virtual outport to every physical channel. If multi-channel sound is provided, raise an error.

  • a single int (example: J)

    This is equivalent to [J]. The first virtual outport will be connected to physical channel J. Note this is NOT the same as ‘mono’, because only one speaker plays, instead of all speakers.

  • a list (example: [I, J])

    The first virtual outport will be connected to physical channel I. The second virtual outport will be connected to physical channel J. And so on. If 1-dimensional sound is provided, play the same to all speakers (like mono mode). If multi-channel sound is provided and the number of channels is different form the length of this list, raise an error.

jack.Client s can’t be kept alive, so this must be called just before processing sample starts.

run()[source]

Start the process, boot the server, start processing frames and wait for the end.

quit()[source]

Set the JackClient.quit_evt

process(frames)[source]

Process a frame of audio.

If the JackClient.play_evt is not set, fill port buffers with zeroes.

Otherwise, pull frames of audio from the JackClient.q until it’s empty.

When it’s empty, set the JackClient.stop_evt and clear the JackClient.play_evt .

Parameters

frames – number of frames (samples) to be processed. unused. passed by jack client

pad(data: numpy.ndarray) numpy.ndarray[source]

Pad a sound that is not a full chunk length long, either by filling with silence or with a continuous sound, if any.

Parameters

data (numpy.ndarray) – The sound to pad!

Returns

numpy.ndarray - the padded sound!

write_to_outports(data)[source]

Write the sound in data to the outport(s).

If self.mono_output:
If data is 1-dimensional:

Write that data to the single outport, which goes to all speakers.

Otherwise, raise an error.

If not self.mono_output:
If data is 1-dimensional:

Write that data to every outport

If data is 2-dimensional:

Write one column to each outport, raising an error if there is a different number of columns than outports.