tgcalls

Voice chats, private incoming and outgoing calls in Telegram for Developers
Examples • Documentation • Channel • Chat

Telegram WebRTC (VoIP)

This project consists of two main parts: tgcalls, pytgcalls. The first is a C++ Python extension. The second uses the extension along with MTProto and provides high level SDK. All together, it allows you to create userbots that can record and broadcast in voice chats, make and receive private calls.

Pyrogram's snippet

from pyrogram import Client, filters
from pyrogram.utils import MAX_CHANNEL_ID

from pytgcalls import GroupCallFactory

app = Client('pytgcalls')
group_call = GroupCallFactory(app).get_file_group_call('input.raw')


@group_call.on_network_status_changed
async def on_network_changed(context, is_connected):
    chat_id = MAX_CHANNEL_ID - context.full_chat.id
    if is_connected:
        await app.send_message(chat_id, 'Successfully joined!')
    else:
        await app.send_message(chat_id, 'Disconnected from voice chat..')


@app.on_message(filters.outgoing & filters.command('join'))
async def join_handler(_, message):
    await group_call.start(message.chat.id)


app.run()

Telethon's snippet

from telethon import TelegramClient, events

from pytgcalls import GroupCallFactory

app = TelegramClient('pytgcalls', api_id, api_hash).start()
group_call_factory = GroupCallFactory(app, GroupCallFactory.MTPROTO_CLIENT_TYPE.TELETHON)
group_call = group_call_factory.get_file_group_call('input.raw')


@app.on(events.NewMessage(outgoing=True, pattern=r'^/join$'))
async def join_handler(event):
    chat = await event.get_chat()
    await group_call.start(chat.id)

app.run_until_disconnected()

Features

• Python solution.
• Supporting popular MTProto libraries: Pyrogram, Telethon.
• Abstract class to implement own MTProto bridge.
• Work with voice chats in channels and chats.
• Multiply voice chats (example).
• System of custom handlers on events.
• Join as channels or chats.
• Join using invite (speaker) links.
• Speaking status with audio levels inside and outside of voice chat.
• Mute/unmute, pause/resume, stop/play, volume control and more...

Available sources of input/output data transfers

• File (GroupCallFile, playout example, recording example) — to use any audio files including named pipe (FIFO).
• Device (GroupCallDevice, example) — to use system virtual devices. Please don't use it with real microphone, headphones, etc.
• Raw (GroupCallRaw, example of restreaming) — to send and receive data in bytes directly from Python.

Note: All audio data is transmitted in PCM 16 bit, 48k. Example how to convert files using FFmpeg.

Requirements

• Python 3.6 or higher.
• A Telegram API key.
• x86_64/arm64 platform and Unix system (use WSL for Windows).

TODO list

• Incoming and Outgoing calls (already there and working, but not in release).
• Private and group video calls.
• Python binary wheels for Windows and more...

Installing

For Pyrogram

pip3 install -U pytgcalls[pyrogram]

For Telethon

pip3 install -U pytgcalls[telethon]

---

PyPi • Sources

tgcalls

The first part of the project is C++ extensions for Python. Pybind11 was used to write it. Binding occurs to the tgcalls library by Telegram, which is used in all official clients. To implement the binding, the code of Telegram Desktop and Telegram Android was studied. Changes have been made to the Telegram library. All modified code is available as a subtree in this repository. The main idea of the changes is to improve the sound quality by making the minimum number of code edits for a simple update. In addition, audio modules for WebRTC were developed. Modules are allowing to transfer audio data directly from Python via bytes, transfer and control the playback/recording of a file or a virtual system device.

How to build

Short answer for linux:

git clone [e-mail]:MarshalX/tgcalls.git --recursive
cd tgcalls

For x86_64:

docker-compose up tgcalls_x86_64

For AArch64 (ARM64):

docker-compose up tgcalls_aarch64

Python wheels will be available in dist folder in root of tgcalls.

More info:

• Manylinux.
• Ubuntu.
• macOS.
• Windows.

Also, you can investigate into manylinux GitHub Actions builds.

Documentation

Temporarily, instead of documentation, you can use an example along with MTProto.

---

Documentation • PyPi • Sources

pytgcalls

This project is implementation of using tgcalls Python binding together with MTProto. By default, this library are supports Pyrogram and Telethon clients for working with Telegram Mobile Protocol. You can write your own implementation of abstract class to work with other libraries.

Learning by example

Visit this page to discover the official examples.

Documentation

pytgcalls's documentation lives at tgcalls.org.

Audio file formats

RAW files are now used. You will have to convert to this format yourself using ffmpeg. The example how to transcode files from a code is available here.

From mp3 to raw (to play in voice chat):

ffmpeg -i input.mp3 -f s16le -ac 2 -ar 48000 -acodec pcm_s16le input.raw

From raw to mp3 (files with recordings):

ffmpeg -f s16le -ac 2 -ar 48000 -acodec pcm_s16le -i output.raw clear_output.mp3

For playout live stream you can use this one:

ffmpeg -y -i http://stream2.cnmns.net/hope-mp3 -f s16le -ac 2 -ar 48000 -acodec pcm_s16le input.raw

For YouTube videos and live streams you can use youtube-dl:

ffmpeg -i "$(youtube-dl -x -g "https://youtu.be/xhXq9BNndhw")" -f s16le -ac 2 -ar 48000 -acodec pcm_s16le input.raw

And set input.raw as input filename.

---

Getting help

You can get help in several ways:

• We have a community of developers helping each other in our Telegram group.
• Report bugs, request new features or ask questions by creating an issue or a discussion.

Contributing

Contributions of all sizes are welcome.

Special thanks to

• @FrayxRulez for amazing code of Unigram.
• @john-preston for Telegram Desktop and tgcalls.
• @bakatrouble for help and inspiration by pytgvoip.
• @delivrance for Pyrogram.
• @Lonami for Telethon.

License

You may copy, distribute and modify the software provided that modifications are described and licensed for free under LGPL-3. Derivatives works (including modifications or anything statically linked to the library) can only be redistributed under LGPL-3, but applications that use the library don't have to be.