Table of Contents

Audio devices

The SDK plays and captures audio through a platform backend (CalloraVoipSdk.Audio.Linux or CalloraVoipSdk.Audio.Windows). The VoipClient exposes enumeration, hot-switch, mute and volume at runtime.

Attach the default device

await client.AttachDefaultAudioAsync(call);
await client.DetachDefaultAudioAsync(call);

With EnableAutomaticAudioDeviceSelection (default true) the SDK picks the OS default input/output. Run headless by leaving the device unattached and using a media tap instead.

Enumerate devices

IReadOnlyList<AudioDeviceDescriptor> inputs  = client.GetAvailableInputAudioDevices();
IReadOnlyList<AudioDeviceDescriptor> outputs = client.GetAvailableOutputAudioDevices();
AudioDeviceRuntimeSnapshot snapshot          = client.GetAudioDeviceRuntimeSnapshot();

The snapshot reports the currently selected devices, mute and volume state.

Hot-switch, mute, volume

These take effect on active calls without renegotiation:

client.SwitchAudioInputDevice(deviceId);
client.SwitchAudioOutputDevice(deviceId);

client.SetAudioInputVolume(0.8f);    // 0.0 – 1.0
client.SetAudioOutputVolume(1.0f);

client.SetAudioInputMuted(true);
client.SetAudioOutputMuted(false);

client.UpdateAudioFormat(new AudioDeviceFormat(/* sample rate, channels … */));

Passing null to SwitchAudioInputDevice/SwitchAudioOutputDevice reverts to the OS default.

Codec preference

Wire format is chosen by SDP negotiation. Bias it with an ordered list once at construction:

new VoipConfiguration
{
    PreferredAudioCodecs = new[] { "OPUS", "G722", "PCMU", "PCMA" }
};

Codec support: the platform audio backends (Audio.Linux / Audio.Windows) decode and encode PCMU, PCMA, G.722 and Opus (RFC 7587, natively at 48 kHz). A call that negotiates any of these plays through AttachDefaultAudioAsync without extra work. Opus is opt-in — list "OPUS" in PreferredAudioCodecs (or it is dropped from the offer). For other/custom codecs, drive the call through a media tap with your own codec instead.

Known limitation (G.722): G.722 transcoding currently resets its ADPCM predictor state per frame instead of carrying it across frames, which can cause audible artifacts. Prefer Opus or G.711 (PCMU/PCMA) where audio quality matters until this is fixed (issue tracker).

Edit this page