RPC command SET_USER_VOICE_SETTINGS "volume" not sending good values

[mthsgl]

Description

When I send a SET_USER_VOICE_SETTINGS command to my RPC application (using pypresence library), the volume value I'm sending is not the one displayed on my discord client.
Valeurs discord.xlsx

Steps to Reproduce

1. Create a application and connect to it in RPC
2. Send a SET_USER_VOICE_SETTINGS command with arg : "volume" = [int between 0 and 200]

Expected Behavior

Sending the value "volume" to my client

Current Behavior

Values not corresponding even though API is answering with the value I sent

Screenshots/Videos

You can find attached an excel file with the value I sent and the one displayed on Discord Client

Client and System Information

Using pypresence and python on a W11 machine

[lsdimagine]

@mthsgl do you mind posting all params in your command?

[mthsgl]

Here is the Payload schema sent by pypresence library :
`def set_user_voice_settings(cls, user_id: str, pan_left: float = None,
pan_right: float = None, volume: int = None,
mute: bool = None):
payload = {
"cmd": "SET_USER_VOICE_SETTINGS",
"args": {
"user_id": str(user_id),
"pan": {
"left": pan_left,
"right": pan_right
},
"volume": volume,
"mute": mute
},
"nonce": '{:.20f}'.format(cls.time())
}

    print (payload)

    return cls(payload, True)`

And here is the payload for a user in my vocal:
{ "cmd": "SET_USER_VOICE_SETTINGS", "args": { "user_id": "933262680758825011", "pan": { "left": null, "right": null }, "volume": 29, "mute": null }, "nonce": "1736411427.93830204010009765625" }

Note: it changed the volume of the user to 78 on my discord client

[rcombs]

This is by-design. The Discord UI exposes a "perceptual" volume level, which has a non-linear relationship with the actual underlying amplitude. The exact function used for the conversion isn't specified, and may change in the future; RPC clients shouldn't rely on it. The APC API exposes the underlying linear amplitude value, so its behavior is guaranteed to remain stable. If exposing this value to a user in a slider or similar UI element, I'd suggest applying your own perceptual curve (a power function with an exponent between 2 and 5 tends to work well; the exact optimal curve depends on your design).