DiscordAudioStream

A simple tool for Windows to stream audio on Discord: stream what you are playing or recording on your computer to a Discord voice channel at higher quality audio via a secondary bot connection.

Video (in French) on how to setup and configure the tool https://www.youtube.com/watch?v=D8GaGncp84o And another one on how to use it https://www.youtube.com/watch?v=Dm1fqSWtyho

Setup

• Get the latest DiscordAudioStream version here https://github.com/eoryl/DiscordAudioStream/releases. Get the installer (DiscordAudioStream_x.y.z_setup.exe) or the standalone package (DiscordAudioStream_x.y.z.zip)
• The installer will propose you to install the needed third party dependencies. If you do not use the installer, make sure you have the minimum versions of .NET and Visual C++ runtimes.
  • DiscordAudioStream and its dependicies require at least .NET 4.7.2 (older versions could run on 4.6.1). If you are running Windows 10 up to date you should have the correct version of .Net runtime. If not or running Windows 7 install at least .NET 4.7.2 runtime or preferably the lastest .NET 4.x framework runtime available for your platform. .NET runtime installers can be found here https://dotnet.microsoft.com/download/dotnet-framework
  • You can get Visual C++ 2019 x86 redist from here https://support.microsoft.com/en-us/help/2977003/the-latest-supported-visual-c-downloads
• Decide which audio device input or ouput you want to stream. To stream what you are playing, the simplest method is to use a virtual audio device. I tested with virtual audio cable https://www.vb-audio.com/Cable/, but you can also use any audio device output with WASAPI loopback capture mode.
• Get a discord bot token at https://discordapp.com/developers/applications (you can find a better in depth tutorial here on how to create a bot https://docs.stillu.cc/guides/getting_started/first-bot.html)
  • Create an application
  • Then in your application go to bot and "Add bot" (tick public if you intend to add it on someone else's server)
  • Add the bot to your server https://discord.com/api/oauth2/authorize?client_id=YOUR_BOT_ID_HERE&permissions=3145728&scope=bot I have tried to limit to necessary permissions only to do audio, just replace YOUR_BOT_ID_HERE with your actual bot client ID
• Run DiscordAudioStream in File > Settings enter your bot token and exit
• Run DiscordAudioStream again the application should connect the status should go to green / idle and the bot connect to the voice channel
• Select your audio device endpoint. If using Virtual Audio Cable and default settings thats "CABLE OUTPUT (VB-Audio Virtual". The meter should move when you play audio via the selected audio device.
• Then select the server and the voice channel, the should status go to streaming.

System level audio configuration

• With MME capture mode (default), all format will be converted automatically. When using WASAPI capture you have to set your card format to 48khz 16bit stereo. For Virtual Audio Cable it is in Settings > Sounds > Sound Control Pannel > Recording > Cable output > Properties > Advanced. You should also use 48khz for playback in Settings > Sounds > Sound Control Pannel > Playback > Cable input > Properties > Advanced, and in your playout application (audio clips and output).For MME this optional but recommended to avoid unecessary resampling and dithering along the chain as Discord expects 48kHz audio.
• Usually your audio applications plays via the Windows default audio device. If you want to stream audio from another device set that device either in the app itself if it allows it otherwise on Windows 10 (1803 and later) you can set which audio device each application uses in Settings > Sounds > Advanced Sound Options > App volume and device preferences. You will get to a pannel that acts as mini audio routing table. Just make sure you use a different one for Discord if you connect to the same channel to avoid feedback).

Application settings

• Audio device: Choose here the audio device input you want to record from or the output you want to clone if using WASAPI loopback mode.
• Server: The Discord server you connect to.
• Voice channel: The voice channel you stream to.
• Bot token: This is your private key for DiscordAudioStream to connect to Discord. Follow the instructions in the Setup section to create one.
• Capture API: This is how the app connects to the driver. MME is the oldest and is more compatible but less performant. WASAPI is more recent and has better performance but often not as well supported by audio device manufacturers. It also does not support sample rate conversion and requires manually setting the device to 48kHz capture sample rate (Discord only support 48KHz and current version of DiscordAudioStream does not embark a sample rate converter). WASAPI shared mode usually works with most configuration to record from inputs (with Virtual Audio Cable for instance). WASAPI exclusive mode is the most performant and stable (should support lowest latencies) but requires exclusive access to the audio input and may simply not work with some audio drivers. WASAPI looback mode does not record from a card input but duplicates the output of a card (so you can do without the virtual audio cable depending on your audio cards and use case).
• Audio API capture buffer: This is how frequently the app will receive audio from the driver. With a lower value you will get a responsive peak meter but you pontetially increase chances of audio dropouts if your CPU gets a bit busy.
• Audio codec content type: That's how Discord.net presents Opus encoding parameters for content. It should be self explanatory, but you can get more details from the Opus codec site.
• Encoder bitrate: This is the audio quality of your stream. 96kbps is the default. 128kbps the highest quality supported.
• Packet loss: This is Opus builtin packet loss management (unsure if this is actually used by Discord)
• Streaming buffer duration: How much audio is cached in the record buffer before sending. Higher is supposedly more tolerant to network degradation/jitter but less responsive.

FAQ

• The app won't connect and what is that bot token

Usually that means you have not set your bot token. See the Setup section for instructions on how to create your bot token. This could also mean your token is already in use by another instance of DiscordAudioStream. Don't run multiple DiscordAudioStream with the same token and don't share your key it's private and should remain secret only known to you.

• The server I want to connect to is not available in the list

Once you have created the bot and its token you must add it to your server and grant it right to list, connect and talk on the voice channel. This means somebody with the "manage server" role must add it to the server. See above in setup

• Which audio capture API should I use?

The most compatible mode is probably MME. The most performant is WASAPI but it requires settings the device sample rate to 48kHz (see above).

• I get an error about sampling rate when I select my card

That's because you are in WASAPI mode. Discord expcets 48kHz audio and there is no real time sample rate conversion in that mode. In Windows go to Settings > Sounds > Sound Control Pannel > Recording|Playback > your card endpoint > Properties > Advanced and set 48khz (ideally 16bit stereo). If your card does't support it switch to MME.

• What are compatibility parameters?

MME capture mode, 100ms capture buffer, Mixed content, 96kbps, 15% packet loss, 1000 ms network buffer.

• What are performance parameters?

WASAPI capture mode, 40ms capture buffer, Music content, 128kbps, 20% packet loss, 200 ms or less network buffer.

• Why can't I go over 128kbps quality?

Discord API won't allow it on regular servers. Non boosted servers are in fact defaulted at 64kbps and capped at 96kbps, but stream via the bot connection can go up to 128kbps. Higher bitrate could be possible. But not test have been done sor far. According to xiph.org tests (https://opus-codec.org/comparison/) after 96kbps you are already past the point of dimnishing returns for the Opus codec anyway.

• Where are the logs?

There are no file based logs at the moment but you can get some logs with debugview (https://docs.microsoft.com/en-us/sysinternals/downloads/debugview).

• The app crashes at startup

Check the setup instructions about prerequisites.

• The app freezes when I connect to a voice channel

Check the setup instructions about prerequisites. More specifically Visual C++ 2019 x86 redist. That's when libsodium and opus.dll are used they are the ones that depend on the VC++ redist.

• But I installed them and it was working before!

Maybe something isn't right in the config. Just delete %APPDATA%\Local\DiscordAudioStream (C:\Users\\\AppData\Local\DiscordAudioStream) and reset your bot token and other params.

• What about supported Operating Systems?

Only tested on up to date Windows 10 64bit. Althought it should work on Windows 10 32bit as the app runs in 32 bit mode. Windows 7 should also be supported as well but is untested.

Compiling

Compilation should work without tinkering using visual studio 2019 with .Net desktop development workload. Note some nugets and their repsective dependencies are required :

• Discord.net: https://github.com/discord-net/Discord.Net
• NAudio: https://github.com/naudio/NAudio

Visual studio should resolve all for you.

After compilation add the dependencies dlls (32bit version) opus.dll and libsodium.dll in the bin/[profile] folder

If you want to recompile them as well instead of using the provided ones, source code is available on their respective website:

• Opus: https://opus-codec.org/
• libsodium: https://doc.libsodium.org/

Alternatives

• The initial idea from Rollyjumper bot in Rust https://github.com/Rollyjumper/fp_music_bot
• A similar tool written in Python https://github.com/QiCuiHub/discord-audio-pipe/