Installation

In short, you can install a known compatible version of ffmpeg and the latest stable version over pip.

pip install python-twitch-stream

Make sure to also install a recent ffmpeg version:

sudo add-apt-repository ppa:mc3man/trusty-media
sudo apt-get update && sudo apt-get install ffmpeg

The ffmpeg library needs to be very recent (written in october 2015). There are plenty of bugs when running a stream using older versions of ffmpeg or avconv, including but not limited to 6GB of memory use, problems with the audio and synchronization of the audio and the video.

Or alternatively, install the latest python-twitch-stream development version via:

pip install git+https://github.com/317070/python-twitch-stream

Documentation

Documentation is available online: http://python-twitch-stream.readthedocs.org/

For support, please use the github issues on the repository.

Example

This is a small example which creates a twitch stream which changes the color of the video according to the colors provided in the chat.

from __future__ import print_function
from twitchstream.outputvideo import TwitchBufferedOutputStream
from twitchstream.chat import TwitchChatStream
import argparse
import time
import numpy as np

if __name__ == "__main__":
    parser = argparse.ArgumentParser(description=__doc__)
    required = parser.add_argument_group('required arguments')
    required.add_argument('-u', '--username',
                          help='twitch username',
                          required=True)
    required.add_argument('-o', '--oauth',
                          help='twitch oauth '
                               '(visit https://twitchapps.com/tmi/ '
                               'to create one for your account)',
                          required=True)
    required.add_argument('-s', '--streamkey',
                          help='twitch streamkey',
                          required=True)
    args = parser.parse_args()

    # load two streams:
    # * one stream to send the video
    # * one stream to interact with the chat
    with TwitchBufferedOutputStream(
            twitch_stream_key=args.streamkey,
            width=640,
            height=480,
            fps=30.,
            enable_audio=True,
            verbose=False) as videostream, \
        TwitchChatStream(
            username=args.username.lower(),  # Must provide a lowercase username.
            oauth=args.oauth,
            verbose=False) as chatstream:

        # Send a chat message to let everybody know you've arrived
        chatstream.send_chat_message("Taking requests!")

        frame = np.zeros((480, 640, 3))
        frequency = 100
        last_phase = 0

        # The main loop to create videos
        while True:

            # Every loop, call to receive messages.
            # This is important, when it is not called,
            # Twitch will automatically log you out.
            # This call is non-blocking.
            received = chatstream.twitch_receive_messages()

            # process all the messages
            if received:
                for chat_message in received:
                    print("Got a message '%s' from %s" % (
                        chat_message['message'],
                        chat_message['username']
                    ))
                    if chat_message['message'] == "red":
                        frame[:, :, :] = np.array(
                            [1, 0, 0])[None, None, :]
                    elif chat_message['message'] == "green":
                        frame[:, :, :] = np.array(
                            [0, 1, 0])[None, None, :]
                    elif chat_message['message'] == "blue":
                        frame[:, :, :] = np.array(
                            [0, 0, 1])[None, None, :]
                    elif chat_message['message'].isdigit():
                        frequency = int(chat_message['message'])

            # If there are not enough video frames left,
            # add some more.
            if videostream.get_video_frame_buffer_state() < 30:
                videostream.send_video_frame(frame)

            # If there are not enough audio fragments left,
            # add some more, but take care to stay in sync with
            # the video! Audio and video buffer separately,
            # so they will go out of sync if the number of video
            # frames does not match the number of audio samples!
            elif videostream.get_audio_buffer_state() < 30:
                x = np.linspace(last_phase,
                                last_phase +
                                frequency*2*np.pi/videostream.fps,
                                int(44100 / videostream.fps) + 1)
                last_phase = x[-1]
                audio = np.sin(x[:-1])
                videostream.send_audio(audio, audio)

            # If nothing is happening, it is okay to sleep for a while
            # and take some pressure of the CPU. But not too long, if
            # the buffers run dry, audio and video will go out of sync.
            else:
                time.sleep(.001)

For a fully-functional example, see examples/color.py, and check the Tutorial for in-depth explanations of the same. More examples are maintained in the examples directory.

Development

Python-twitch-stream is a work in progress, but is stable. Feel free to ask for features or add pull-requests with updates on the code.

Changelog