NOTE: pymumble is currently being reworked from a stateful application object to a library that can be used both as a traditional client and as an independent collection of mostly stateless functionality. The documentation is kept more or less up to date, but until 2.0.0 is released, nothing is stable.

pymumble - Python Mumble client library

Installation

Requirements

opus is required to send and receive audio.

Install with uv or pip

One of the following:

• uv add pymumble
• uv pip install pymumble
• pip install pymumble

Install from git

git clone https://codeberg.org/pymumble/pymumble.git
cd pymumble
uv sync

Usage

# /// script
# dependencies = [
#     "pymumble>=2",
# ]
# ///

from mumble import Mumble
from subprocess import PIPE, Popen

with Mumble("127.0.0.1", "A Weedy Samaritan", debug=False) as m:
    usernames = ", ".join(u["name"] for u in m.my_channel().get_users())
    m.my_channel().send_text_message(
        f"Hello, {usernames}. You're all Brian! You're all individuals!"
    )

    sound = Popen("ffmpeg -i audio.opus -ar 48000 -b:a 64k -f s16le -".split(), stdout=PIPE).stdout.read()
    m.send_audio.add_sound(sound).wait()

BREAKING CHANGES and updates in pymumble 2.0.0

The following enhancements are included in pymumble 2.0.0:

• Implement encrypted UDP audio & pings with AES-OCB2, compatible with the latest Mumble server.
• Implement unencrypted UDP pings to retrieve extended server info before connecting.
• Support the latest protocol version: 1.5.735.
• Use uv for packaging and ruff for linting.
• Send functional version string compatible with the latest Mumble server.
• Support Python with statements.

In order to bring pymumble up to date with modern python development practices, the following breaking changes have been introduced in version 2.0.0:

• Change the import path from pymumble_py3 to simply mumble.
• Replace callback constants with a class interface. e.g., Mumble.callbacks.set_callback(PYMUMBLE_CLBK_SOUNDRECEIVED, ...) is now Mumble.callbacks.sound_received.set_handler(...).
• Remove Mumble.set_receive_sound(), audio support is now enabled by default.
  • To disable audio support and avoid importing opuslib, instantiate the Mumble object with Mumble(enable_audio=False).
  • To disable audio support after the object has been created set m = Mumble(); m.enable_audio = False. This will still import opuslib.
• The Mumble class getter/setter functions set_application_string(), set_loop_rate(), and get_loop_rate() have been removed. These parameters can be set in the Mumble object initializer and queried and changed by accessing the object's application and loop_rate public attributes.
• Drop support for legacy audio codecs.
• Rename all constants to drop the PYMUMBLE_ prefix, some also renamed for clarity.
• Rename SoundOutput to SendAudio and SoundQueue to ReceivedAudioQueue.

Because pymumble now follows the Semantic Versioning 2.0.0 versioning scheme, any further breaking changes must increment the MAJOR version number. The dependency

pymumble>=2,<3

will resolve to the latest version with a stable API.

Goals

We strive to keep pymumble up to date with the latest version of the mumble protocol and compatible with the mumble server. Pymumble should be bit-compatible with the official mumble client at the application/messaging layer. (The python & C++ protobuf implementations may vary in how they serialise the data over the wire, which is acceptable per the protocol buffer specification.)

Future releases will focus on:

• improving the API
• improving performance
• improving test coverage
• improving documentation
• rounding out protocol support

Contributing guidelines

• Follow conventional commit guidelines for writing commit messages.
• Document code in the Sphinx docstring format.
• Include integration tests for updated or new functionality.
• Format code with ruff format and lint with ruff lint before committing.

Applications using pymumble

• Abot
• Botamusique
• MumbleRadioPlayer (archived)
• MumbleRecbot (deprecated)

Thanks

• @azlux for maintaining the pymumble library before version 2
• Jan Petykiewicz for the AES-OCB2 implementation
• Ranomier for the python3 port
• @raylu for making pymumble speak into channels
• @schlarpc for fixes on buffer
• @Robert904 for the inital pymumble implementation
• All contributors to the previous versions.