diff --git a/examples/simple_room/src/app.rs b/examples/simple_room/src/app.rs
index ab80c0f..0cdd06a 100644
--- a/examples/simple_room/src/app.rs
+++ b/examples/simple_room/src/app.rs
@@ -1,12 +1,15 @@
 use crate::events::UiCmd;
 use crate::logo_track::LogoTrack;
+use crate::sine_track::SineTrack;
 use crate::video_renderer::VideoRenderer;
 use crate::{events::AsyncCmd, video_grid::VideoGrid};
 use egui::{Rounding, Stroke};
 use egui_wgpu::WgpuConfiguration;
+use futures::StreamExt;
 use image::ImageFormat;
 use livekit::options::{TrackPublishOptions, VideoCaptureOptions};
 use livekit::prelude::*;
+use livekit::webrtc::audio_stream::native::NativeAudioStream;
 use livekit::webrtc::native::yuv_helper;
 use livekit::webrtc::video_frame::native::I420BufferExt;
 use livekit::webrtc::video_frame::{I420Buffer, VideoFrame, VideoRotation};
@@ -36,6 +39,7 @@ use winit::{
 struct Session {
     room: Room,
     logo_track: LogoTrack,
+    sine_track: SineTrack,
     close_tx: oneshot::Sender<()>,
     handle: tokio::task::JoinHandle<()>,
 }
@@ -113,6 +117,7 @@ pub fn run(rt: tokio::runtime::Runtime) {
                         if let Ok((room, room_events)) = res {
                             let (close_tx, close_rx) = oneshot::channel();
                             let logo_track = LogoTrack::new(room.session());
+                            let sine_track = SineTrack::new(room.session());
                             let handle = tokio::spawn(room_task(
                                 state.clone(),
                                 room_events,
@@ -123,6 +128,7 @@ pub fn run(rt: tokio::runtime::Runtime) {
                             *state.session.lock() = Some(Session {
                                 room,
                                 logo_track,
+                                sine_track,
                                 close_tx,
                                 handle,
                             });
@@ -156,6 +162,12 @@ pub fn run(rt: tokio::runtime::Runtime) {
                             }
                         }
                     }
+                    AsyncCmd::ToggleSine => {
+                        if let Some(session) = state.session.lock().as_mut() {
+                            let sine_track = &mut session.sine_track;
+                            sine_track.publish().await.unwrap();
+                        }
+                    }
                 }
             }
         });
@@ -213,8 +225,15 @@ impl App {
                                     self.video_renderers
                                         .insert((participant.sid(), track.sid()), video_renderer);
                                 }
-                                RemoteTrack::Audio(_) => {
-                                    // The demo doesn't support Audio rendering at the moment.
+                                RemoteTrack::Audio(audio_track) => {
+                                    tokio::spawn(async move {
+                                        let mut stream =
+                                            NativeAudioStream::new(audio_track.rtc_track());
+
+                                        while let Some(_frame) = stream.next().await {
+                                            // Received audio frames
+                                        }
+                                    });
                                 }
                             };
                         }
@@ -323,9 +342,12 @@ impl App {
                 });
 
                 ui.menu_button("Publish", |ui| {
-                    if ui.button("CustomTrack - LK Logo").clicked() {
+                    if ui.button("Logo").clicked() {
                         let _ = self.cmd_tx.send(AsyncCmd::ToggleLogo);
                     }
+                    if ui.button("SineWave").clicked() {
+                        let _ = self.cmd_tx.send(AsyncCmd::ToggleSine);
+                    }
                 });
             });
         });
diff --git a/examples/simple_room/src/events.rs b/examples/simple_room/src/events.rs
index 46bd655..34cf9b3 100644
--- a/examples/simple_room/src/events.rs
+++ b/examples/simple_room/src/events.rs
@@ -7,6 +7,7 @@ pub enum AsyncCmd {
     RoomDisconnect,
     SimulateScenario { scenario: SimulateScenario },
     ToggleLogo, // Unpublish/Publish a logo track
+    ToggleSine,
 }
 
 #[derive(Debug)]
diff --git a/examples/simple_room/src/logo_track.rs b/examples/simple_room/src/logo_track.rs
index 6dd802b..46c5c6a 100644
--- a/examples/simple_room/src/logo_track.rs
+++ b/examples/simple_room/src/logo_track.rs
@@ -56,7 +56,7 @@ impl LogoTrack {
     }
 
     pub async fn publish(&mut self) -> Result<(), RoomError> {
-        self.unpublish().await;
+        self.unpublish().await?;
 
         let (close_tx, close_rx) = oneshot::channel();
         let track = LocalVideoTrack::create_video_track(
diff --git a/examples/simple_room/src/main.rs b/examples/simple_room/src/main.rs
index 1e87b11..7d2533c 100644
--- a/examples/simple_room/src/main.rs
+++ b/examples/simple_room/src/main.rs
@@ -1,6 +1,7 @@
 mod app;
 mod events;
 mod logo_track;
+mod sine_track;
 mod video_grid;
 mod video_renderer;
 
diff --git a/examples/simple_room/src/sine_track.rs b/examples/simple_room/src/sine_track.rs
new file mode 100644
index 0000000..43231a9
--- /dev/null
+++ b/examples/simple_room/src/sine_track.rs
@@ -0,0 +1,127 @@
+use livekit::options::{AudioCaptureOptions, TrackPublishOptions};
+use livekit::webrtc::audio_frame::AudioFrame;
+use livekit::{prelude::*, webrtc::audio_source::native::NativeAudioSource};
+use parking_lot::Mutex;
+use std::sync::Arc;
+use std::time::Duration;
+use tokio::sync::oneshot;
+use tokio::task::JoinHandle;
+
+#[derive(Clone)]
+struct FrameData {
+    pub sample_rate: u32,
+    pub freq: f64,
+    pub amplitude: f64,
+}
+
+impl Default for FrameData {
+    fn default() -> Self {
+        Self {
+            sample_rate: 48000,
+            freq: 440.0,
+            amplitude: 1.0,
+        }
+    }
+}
+
+struct TrackHandle {
+    frame_data: Arc<Mutex<FrameData>>,
+    close_tx: oneshot::Sender<()>,
+    track: LocalAudioTrack,
+    task: JoinHandle<()>,
+}
+
+pub struct SineTrack {
+    rtc_source: NativeAudioSource,
+    session: RoomSession,
+    handle: Option<TrackHandle>,
+}
+
+impl SineTrack {
+    pub fn new(session: RoomSession) -> Self {
+        Self {
+            rtc_source: NativeAudioSource::default(),
+            session,
+            handle: None,
+        }
+    }
+
+    pub async fn publish(&mut self) -> Result<(), RoomError> {
+        let (close_tx, close_rx) = oneshot::channel();
+        let track = LocalAudioTrack::create_audio_track(
+            "sine_wave",
+            AudioCaptureOptions {
+                auto_gain_control: false,
+                echo_cancellation: false,
+                noise_suppression: false,
+            },
+            self.rtc_source.clone(),
+        );
+
+        let data = Arc::new(Mutex::new(FrameData::default()));
+        let task = tokio::spawn(Self::track_task(
+            close_rx,
+            self.rtc_source.clone(),
+            data.clone(),
+        ));
+
+        self.session
+            .local_participant()
+            .publish_track(
+                LocalTrack::Audio(track.clone()),
+                TrackPublishOptions {
+                    source: TrackSource::Microphone,
+                    ..Default::default()
+                },
+            )
+            .await?;
+
+        let handle = TrackHandle {
+            frame_data: data,
+            close_tx,
+            track,
+            task,
+        };
+
+        self.handle = Some(handle);
+        Ok(())
+    }
+
+    async fn track_task(
+        mut close_rx: oneshot::Receiver<()>,
+        rtc_source: NativeAudioSource,
+        frame_options: Arc<Mutex<FrameData>>,
+    ) {
+        let mut interval = tokio::time::interval(Duration::from_millis(10));
+        let mut samples_10ms = Vec::<i16>::new();
+
+        loop {
+            interval.tick().await;
+
+            let data = frame_options.lock();
+            let samples_count_10ms = (data.sample_rate / 100) as usize;
+
+            if samples_10ms.capacity() != samples_count_10ms {
+                samples_10ms.resize(samples_count_10ms, 0i16);
+            }
+
+            for i in 0..samples_count_10ms {
+                let val = data.amplitude
+                    * f64::sin(
+                        std::f64::consts::PI * 2.0 * data.freq * i as f64
+                            / samples_count_10ms as f64,
+                    );
+
+                // WebRTC uses 16-bit signed PCM
+                samples_10ms[i] = (val * 32768.0) as i16;
+            }
+
+            rtc_source.capture_frame(AudioFrame {
+                data: samples_10ms.clone(),
+                sample_rate_hz: data.sample_rate,
+                num_channels: 1,
+                samples_per_channel: samples_count_10ms,
+            });
+        }
+    }
+}
diff --git a/livekit-webrtc/src/audio_frame.rs b/livekit-webrtc/src/audio_frame.rs
new file mode 100644
index 0000000..1a86f21
--- /dev/null
+++ b/livekit-webrtc/src/audio_frame.rs
@@ -0,0 +1,7 @@
+#[derive(Debug, Clone)]
+pub struct AudioFrame {
+    pub data: Vec<i16>,
+    pub sample_rate_hz: u32,
+    pub num_channels: usize,
+    pub samples_per_channel: usize,
+}
diff --git a/livekit-webrtc/src/audio_source.rs b/livekit-webrtc/src/audio_source.rs
new file mode 100644
index 0000000..a1ecec3
--- /dev/null
+++ b/livekit-webrtc/src/audio_source.rs
@@ -0,0 +1,25 @@
+use crate::imp::audio_source as imp_as;
+
+#[cfg(not(target_arch = "wasm32"))]
+pub mod native {
+    use super::imp_as;
+    use crate::audio_frame::AudioFrame;
+    use std::fmt::{Debug, Formatter};
+
+    #[derive(Default, Clone)]
+    pub struct NativeAudioSource {
+        pub(crate) handle: imp_as::NativeAudioSource,
+    }
+
+    impl Debug for NativeAudioSource {
+        fn fmt(&self, f: &mut Formatter) -> std::fmt::Result {
+            f.debug_struct("NativeAudioSource").finish()
+        }
+    }
+
+    impl NativeAudioSource {
+        pub fn capture_frame(&self, frame: AudioFrame) {
+            self.handle.capture_frame(frame)
+        }
+    }
+}
diff --git a/livekit-webrtc/src/audio_stream.rs b/livekit-webrtc/src/audio_stream.rs
new file mode 100644
index 0000000..8feeefa
--- /dev/null
+++ b/livekit-webrtc/src/audio_stream.rs
@@ -0,0 +1,48 @@
+use crate::imp::audio_stream as stream_imp;
+
+#[cfg(not(target_arch = "wasm32"))]
+pub mod native {
+    use super::stream_imp;
+    use crate::audio_frame::AudioFrame;
+    use crate::media_stream::RtcAudioTrack;
+    use futures::stream::Stream;
+    use std::fmt::{Debug, Formatter};
+    use std::pin::Pin;
+    use std::task::{Context, Poll};
+
+    pub struct NativeAudioStream {
+        pub(crate) handle: stream_imp::NativeAudioStream,
+    }
+
+    impl Debug for NativeAudioStream {
+        fn fmt(&self, f: &mut Formatter) -> std::fmt::Result {
+            f.debug_struct("NativeAudioStream")
+                .field("track", &self.track())
+                .finish()
+        }
+    }
+
+    impl NativeAudioStream {
+        pub fn new(audio_track: RtcAudioTrack) -> Self {
+            Self {
+                handle: stream_imp::NativeAudioStream::new(audio_track),
+            }
+        }
+
+        pub fn track(&self) -> RtcAudioTrack {
+            self.handle.track()
+        }
+
+        pub fn close(&mut self) {
+            self.handle.close()
+        }
+    }
+
+    impl Stream for NativeAudioStream {
+        type Item = AudioFrame;
+
+        fn poll_next(self: Pin<&mut Self>, cx: &mut Context) -> Poll<Option<Self::Item>> {
+            Pin::new(&mut self.get_mut().handle).poll_next(cx)
+        }
+    }
+}
diff --git a/livekit-webrtc/src/lib.rs b/livekit-webrtc/src/lib.rs
index 269ec6e..ee758f6 100644
--- a/livekit-webrtc/src/lib.rs
+++ b/livekit-webrtc/src/lib.rs
@@ -26,6 +26,9 @@ pub struct RtcError {
     pub message: String,
 }
 
+pub mod audio_frame;
+pub mod audio_source;
+pub mod audio_stream;
 pub mod data_channel;
 pub mod ice_candidate;
 pub mod media_stream;
diff --git a/livekit-webrtc/src/native/audio_source.rs b/livekit-webrtc/src/native/audio_source.rs
new file mode 100644
index 0000000..37af08d
--- /dev/null
+++ b/livekit-webrtc/src/native/audio_source.rs
@@ -0,0 +1,34 @@
+use crate::audio_frame::AudioFrame;
+use cxx::SharedPtr;
+use webrtc_sys::media_stream as sys_ms;
+
+#[derive(Clone)]
+pub struct NativeAudioSource {
+    sys_handle: SharedPtr<sys_ms::ffi::AudioTrackSource>,
+}
+
+impl Default for NativeAudioSource {
+    fn default() -> Self {
+        Self {
+            sys_handle: sys_ms::ffi::new_audio_track_source(),
+        }
+    }
+}
+
+impl NativeAudioSource {
+    pub fn sys_handle(&self) -> SharedPtr<sys_ms::ffi::AudioTrackSource> {
+        self.sys_handle.clone()
+    }
+
+    pub fn capture_frame(&self, frame: AudioFrame) {
+        // TODO(theomonnom): Should we check for 10ms worth of data here?
+        unsafe {
+            self.sys_handle.on_captured_frame(
+                frame.data.as_ptr(),
+                frame.sample_rate_hz as i32,
+                frame.num_channels,
+                frame.samples_per_channel,
+            )
+        }
+    }
+}
diff --git a/livekit-webrtc/src/native/audio_stream.rs b/livekit-webrtc/src/native/audio_stream.rs
new file mode 100644
index 0000000..03503f2
--- /dev/null
+++ b/livekit-webrtc/src/native/audio_stream.rs
@@ -0,0 +1,80 @@
+use crate::{audio_frame::AudioFrame, media_stream::RtcAudioTrack};
+use cxx::UniquePtr;
+use futures::stream::Stream;
+use std::pin::Pin;
+use std::task::{Context, Poll};
+use tokio::sync::mpsc;
+use webrtc_sys::media_stream as sys_ms;
+
+pub struct NativeAudioStream {
+    native_observer: UniquePtr<sys_ms::ffi::NativeAudioSink>,
+    _observer: Box<AudioTrackObserver>,
+    audio_track: RtcAudioTrack,
+    frame_rx: mpsc::UnboundedReceiver<AudioFrame>,
+}
+
+impl NativeAudioStream {
+    pub fn new(audio_track: RtcAudioTrack) -> Self {
+        let (frame_tx, frame_rx) = mpsc::unbounded_channel();
+        let mut observer = Box::new(AudioTrackObserver { frame_tx });
+        let mut native_observer = unsafe {
+            sys_ms::ffi::new_native_audio_sink(Box::new(sys_ms::AudioSinkWrapper::new(
+                &mut *observer,
+            )))
+        };
+
+        unsafe {
+            sys_ms::ffi::media_to_audio(audio_track.sys_handle())
+                .add_sink(native_observer.pin_mut());
+        }
+
+        Self {
+            native_observer,
+            _observer: observer,
+            audio_track,
+            frame_rx,
+        }
+    }
+
+    pub fn track(&self) -> RtcAudioTrack {
+        self.audio_track.clone()
+    }
+
+    pub fn close(&mut self) {
+        self.frame_rx.close();
+        unsafe {
+            sys_ms::ffi::media_to_audio(self.audio_track.sys_handle())
+                .remove_sink(self.native_observer.pin_mut());
+        }
+    }
+}
+
+impl Drop for NativeAudioStream {
+    fn drop(&mut self) {
+        self.close();
+    }
+}
+
+impl Stream for NativeAudioStream {
+    type Item = AudioFrame;
+
+    fn poll_next(mut self: Pin<&mut Self>, cx: &mut Context) -> Poll<Option<Self::Item>> {
+        self.frame_rx.poll_recv(cx)
+    }
+}
+
+pub struct AudioTrackObserver {
+    frame_tx: mpsc::UnboundedSender<AudioFrame>,
+}
+
+impl sys_ms::AudioSink for AudioTrackObserver {
+    fn on_data(&self, data: &[i16], sample_rate: i32, nb_channels: usize, nb_frames: usize) {
+        // TODO(theomonnom): Should we avoid copy here?
+        let _ = self.frame_tx.send(AudioFrame {
+            data: data.to_owned(),
+            sample_rate_hz: sample_rate as u32,
+            num_channels: nb_channels,
+            samples_per_channel: nb_frames,
+        });
+    }
+}
diff --git a/livekit-webrtc/src/native/mod.rs b/livekit-webrtc/src/native/mod.rs
index e890949..6c5f63a 100644
--- a/livekit-webrtc/src/native/mod.rs
+++ b/livekit-webrtc/src/native/mod.rs
@@ -1,3 +1,5 @@
+pub mod audio_source;
+pub mod audio_stream;
 pub mod data_channel;
 pub mod ice_candidate;
 pub mod media_stream;
diff --git a/livekit-webrtc/src/native/peer_connection_factory.rs b/livekit-webrtc/src/native/peer_connection_factory.rs
index 13440a9..a4a2fac 100644
--- a/livekit-webrtc/src/native/peer_connection_factory.rs
+++ b/livekit-webrtc/src/native/peer_connection_factory.rs
@@ -1,6 +1,7 @@
+use crate::audio_source::native::NativeAudioSource;
 use crate::imp::media_stream as imp_ms;
 use crate::imp::peer_connection as imp_pc;
-use crate::media_stream::RtcVideoTrack;
+use crate::media_stream::{RtcAudioTrack, RtcVideoTrack};
 use crate::peer_connection::PeerConnection;
 use crate::peer_connection_factory::{
     ContinualGatheringPolicy, IceServer, IceTransportsType, RtcConfiguration,
@@ -132,6 +133,16 @@ impl PeerConnectionFactory {
         }
     }
 
+    pub fn create_audio_track(&self, label: &str, source: NativeAudioSource) -> RtcAudioTrack {
+        RtcAudioTrack {
+            handle: imp_ms::RtcAudioTrack {
+                sys_handle: self
+                    .sys_handle
+                    .create_audio_track(label.to_string(), source.handle.sys_handle()),
+            },
+        }
+    }
+
     pub fn get_rtp_sender_capabilities(&self, media_type: MediaType) -> RtpCapabilities {
         self.sys_handle
             .get_rtp_sender_capabilities(media_type.into())
diff --git a/livekit-webrtc/src/native/rtp_transceiver.rs b/livekit-webrtc/src/native/rtp_transceiver.rs
index a3119d5..c3d1a8c 100644
--- a/livekit-webrtc/src/native/rtp_transceiver.rs
+++ b/livekit-webrtc/src/native/rtp_transceiver.rs
@@ -19,6 +19,7 @@ impl From<sys_webrtc::ffi::RtpTransceiverDirection> for RtpTransceiverDirection
             sys_webrtc::ffi::RtpTransceiverDirection::SendOnly => Self::SendOnly,
             sys_webrtc::ffi::RtpTransceiverDirection::RecvOnly => Self::RecvOnly,
             sys_webrtc::ffi::RtpTransceiverDirection::Inactive => Self::Inactive,
+            sys_webrtc::ffi::RtpTransceiverDirection::Stopped => Self::Stopped,
             _ => panic!("unknown RtpTransceiverDirection"),
         }
     }
@@ -31,6 +32,7 @@ impl From<RtpTransceiverDirection> for sys_webrtc::ffi::RtpTransceiverDirection
             RtpTransceiverDirection::SendOnly => Self::SendOnly,
             RtpTransceiverDirection::RecvOnly => Self::RecvOnly,
             RtpTransceiverDirection::Inactive => Self::Inactive,
+            RtpTransceiverDirection::Stopped => Self::Stopped,
             _ => panic!("unknown RtpTransceiverDirection"),
         }
     }
diff --git a/livekit-webrtc/src/peer_connection_factory.rs b/livekit-webrtc/src/peer_connection_factory.rs
index 7fedb86..6d1c6c5 100644
--- a/livekit-webrtc/src/peer_connection_factory.rs
+++ b/livekit-webrtc/src/peer_connection_factory.rs
@@ -63,16 +63,22 @@ impl PeerConnectionFactory {
 
 pub mod native {
     use super::PeerConnectionFactory;
-    use crate::media_stream::RtcVideoTrack;
+    use crate::audio_source::native::NativeAudioSource;
+    use crate::media_stream::{RtcAudioTrack, RtcVideoTrack};
     use crate::video_source::native::NativeVideoSource;
 
     pub trait PeerConnectionFactoryExt {
         fn create_video_track(&self, label: &str, source: NativeVideoSource) -> RtcVideoTrack;
+        fn create_audio_track(&self, label: &str, source: NativeAudioSource) -> RtcAudioTrack;
     }
 
     impl PeerConnectionFactoryExt for PeerConnectionFactory {
         fn create_video_track(&self, label: &str, source: NativeVideoSource) -> RtcVideoTrack {
             self.handle.create_video_track(label, source)
         }
+
+        fn create_audio_track(&self, label: &str, source: NativeAudioSource) -> RtcAudioTrack {
+            self.handle.create_audio_track(label, source)
+        }
     }
 }
diff --git a/livekit-webrtc/src/prelude.rs b/livekit-webrtc/src/prelude.rs
index d2d62d3..08fe57a 100644
--- a/livekit-webrtc/src/prelude.rs
+++ b/livekit-webrtc/src/prelude.rs
@@ -1,3 +1,4 @@
+pub use crate::audio_frame::AudioFrame;
 pub use crate::data_channel::{
     DataBuffer, DataChannel, DataChannelError, DataChannelInit, DataState,
 };
diff --git a/livekit/src/room/options.rs b/livekit/src/room/options.rs
index 118810a..8410488 100644
--- a/livekit/src/room/options.rs
+++ b/livekit/src/room/options.rs
@@ -40,38 +40,63 @@ pub struct VideoPreset {
     pub height: u32,
 }
 
+#[derive(Debug, Clone)]
+pub struct AudioEncoding {
+    pub max_bitrate: u64,
+}
+
 #[derive(Debug, Clone)]
 pub struct AudioPreset {
-    pub max_bitrate: u32,
+    pub encoding: AudioEncoding,
 }
 
 impl AudioPreset {
-    pub const fn new(max_bitrate: u32) -> Self {
-        Self { max_bitrate }
+    pub const fn new(max_bitrate: u64) -> Self {
+        Self {
+            encoding: AudioEncoding { max_bitrate },
+        }
+    }
+}
+
+#[derive(Debug, Clone)]
+pub struct AudioCaptureOptions {
+    pub echo_cancellation: bool,
+    pub noise_suppression: bool,
+    pub auto_gain_control: bool,
+}
+
+impl Default for AudioCaptureOptions {
+    fn default() -> Self {
+        Self {
+            echo_cancellation: true,
+            noise_suppression: true,
+            auto_gain_control: true,
+        }
     }
 }
 
 #[derive(Clone, Debug)]
 pub struct VideoCaptureOptions {
-    pub preset: VideoPreset,
+    pub resolution: VideoResolution,
 }
 
 impl Default for VideoCaptureOptions {
     fn default() -> Self {
         Self {
-            preset: video::H720,
+            resolution: video::H720.resolution(),
         }
     }
 }
 
 #[derive(Clone, Debug)]
 pub struct TrackPublishOptions {
-    pub dynacast: bool,
+    // If the encodings aren't set, LiveKit will compute the most appropriate ones
+    pub video_encoding: Option<VideoEncoding>,
+    pub audio_encoding: Option<AudioEncoding>,
     pub video_codec: VideoCodec,
     pub dtx: bool,
     pub red: bool,
     pub simulcast: bool,
-    pub screenshare: bool,
     pub name: String,
     pub source: TrackSource,
 }
@@ -79,12 +104,12 @@ pub struct TrackPublishOptions {
 impl Default for TrackPublishOptions {
     fn default() -> Self {
         Self {
-            dynacast: false,
+            video_encoding: None,
+            audio_encoding: None,
             video_codec: VideoCodec::VP8,
             dtx: true,
             red: true,
             simulcast: true,
-            screenshare: false,
             name: "unnamed track".to_owned(),
             source: TrackSource::Unknown,
         }
@@ -120,7 +145,8 @@ pub fn compute_video_encodings(
     height: u32,
     options: &TrackPublishOptions,
 ) -> Vec<RtpEncodingParameters> {
-    let encoding = compute_appropriate_encoding(options.screenshare, width, height);
+    let screenshare = options.source == TrackSource::Screenshare;
+    let encoding = compute_appropriate_encoding(screenshare, width, height);
 
     let initial_preset = VideoPreset {
         width,
@@ -135,8 +161,7 @@ pub fn compute_video_encodings(
         return into_rtp_encodings(width, height, &[initial_preset]);
     }
 
-    let mut simulcast_presets =
-        compute_default_simulcast_presets(options.screenshare, &initial_preset);
+    let mut simulcast_presets = compute_default_simulcast_presets(screenshare, &initial_preset);
 
     let mid_preset = simulcast_presets.pop();
     let low_preset = simulcast_presets.pop();
diff --git a/livekit/src/room/participant/local_participant.rs b/livekit/src/room/participant/local_participant.rs
index fceb405..a93c567 100644
--- a/livekit/src/room/participant/local_participant.rs
+++ b/livekit/src/room/participant/local_participant.rs
@@ -1,10 +1,12 @@
 use super::{ConnectionQuality, ParticipantInner};
+use crate::options;
 use crate::options::compute_video_encodings;
 use crate::options::video_layers_from_encodings;
 use crate::options::TrackPublishOptions;
 use crate::prelude::*;
 use crate::proto;
 use crate::rtc_engine::RtcEngine;
+use livekit_webrtc::rtp_parameters::RtpEncodingParameters;
 use parking_lot::RwLockReadGuard;
 use std::collections::HashMap;
 use std::sync::Arc;
@@ -53,13 +55,24 @@ impl LocalParticipant {
                 // Get the video dimension
                 // TODO(theomonnom): Use MediaStreamTrack::getSettings() on web
                 let capture_options = video_track.capture_options();
-                req.width = capture_options.preset.width;
-                req.height = capture_options.preset.height;
+                req.width = capture_options.resolution.width;
+                req.height = capture_options.resolution.height;
 
                 encodings = compute_video_encodings(req.width, req.height, &options);
                 req.layers = video_layers_from_encodings(req.width, req.height, &encodings);
             }
-            LocalTrack::Audio(_audio_track) => {}
+            LocalTrack::Audio(_audio_track) => {
+                // Setup audio encoding
+                let audio_encoding = options
+                    .audio_encoding
+                    .as_ref()
+                    .unwrap_or(&options::audio::SPEECH.encoding);
+
+                encodings.push(RtpEncodingParameters {
+                    max_bitrate: Some(audio_encoding.max_bitrate),
+                    ..Default::default()
+                });
+            }
         }
 
         let track_info = self.rtc_engine.add_track(req).await?;
diff --git a/livekit/src/room/track/local_audio_track.rs b/livekit/src/room/track/local_audio_track.rs
index 09f08d2..e55c3d6 100644
--- a/livekit/src/room/track/local_audio_track.rs
+++ b/livekit/src/room/track/local_audio_track.rs
@@ -1,79 +1,100 @@
 use super::TrackInner;
+use crate::options::AudioCaptureOptions;
 use crate::prelude::*;
 use crate::proto;
+use crate::rtc_engine::lk_runtime::LkRuntime;
+use crate::webrtc::peer_connection_factory::native::PeerConnectionFactoryExt;
 use livekit_webrtc as rtc;
+use parking_lot::Mutex;
+use rtc::audio_source::native::NativeAudioSource;
 use std::sync::Arc;
 use tokio::sync::mpsc;
 
+#[derive(Debug)]
+pub struct LocalAudioTrackInner {
+    track_inner: TrackInner,
+    capture_options: Mutex<AudioCaptureOptions>,
+}
+
 #[derive(Clone, Debug)]
 pub struct LocalAudioTrack {
-    pub(crate) inner: Arc<TrackInner>,
+    inner: Arc<LocalAudioTrackInner>,
 }
 
 impl LocalAudioTrack {
     pub(crate) fn new(
-        sid: TrackSid,
         name: String,
         rtc_track: rtc::media_stream::RtcAudioTrack,
+        capture_options: AudioCaptureOptions,
     ) -> Self {
         Self {
-            inner: Arc::new(TrackInner::new(
-                sid,
-                name,
-                TrackKind::Audio,
-                rtc::media_stream::MediaStreamTrack::Audio(rtc_track),
-            )),
+            inner: Arc::new(LocalAudioTrackInner {
+                track_inner: TrackInner::new(
+                    "unknown".to_string().into(), // sid
+                    name,
+                    TrackKind::Audio,
+                    rtc::media_stream::MediaStreamTrack::Audio(rtc_track),
+                ),
+                capture_options: Mutex::new(capture_options),
+            }),
         }
     }
 
+    #[inline]
+    pub fn capture_options(&self) -> AudioCaptureOptions {
+        self.inner.capture_options.lock().clone()
+    }
+
     #[inline]
     pub fn sid(&self) -> TrackSid {
-        self.inner.sid()
+        self.inner.track_inner.sid()
     }
 
     #[inline]
     pub fn name(&self) -> String {
-        self.inner.name()
+        self.inner.track_inner.name()
     }
 
     #[inline]
     pub fn kind(&self) -> TrackKind {
-        self.inner.kind()
+        self.inner.track_inner.kind()
     }
 
     #[inline]
     pub fn source(&self) -> TrackSource {
-        self.inner.source()
+        self.inner.track_inner.source()
     }
 
     #[inline]
     pub fn stream_state(&self) -> StreamState {
-        self.inner.stream_state()
+        self.inner.track_inner.stream_state()
     }
 
     #[inline]
     pub fn start(&self) {
-        self.inner.start()
+        self.inner.track_inner.start()
     }
 
     #[inline]
     pub fn stop(&self) {
-        self.inner.stop()
+        self.inner.track_inner.stop()
     }
 
     #[inline]
     pub fn muted(&self) -> bool {
-        self.inner.muted()
+        self.inner.track_inner.muted()
     }
 
     #[inline]
     pub fn set_muted(&self, muted: bool) {
-        self.inner.set_muted(muted)
+        self.inner.track_inner.set_muted(muted)
     }
 
     #[inline]
     pub fn rtc_track(&self) -> rtc::media_stream::RtcAudioTrack {
-        if let rtc::media_stream::MediaStreamTrack::Audio(audio) = self.inner.rtc_track() {
+        if let rtc::media_stream::MediaStreamTrack::Audio(audio) =
+            self.inner.track_inner.rtc_track()
+        {
             audio
         } else {
             unreachable!()
@@ -82,12 +103,12 @@ impl LocalAudioTrack {
 
     #[inline]
     pub fn register_observer(&self) -> mpsc::UnboundedReceiver<TrackEvent> {
-        self.inner.register_observer()
+        self.inner.track_inner.register_observer()
     }
 
     #[inline]
     pub(crate) fn transceiver(&self) -> Option<rtc::rtp_transceiver::RtpTransceiver> {
-        self.inner.transceiver()
+        self.inner.track_inner.transceiver()
     }
 
     #[inline]
@@ -95,11 +116,25 @@ impl LocalAudioTrack {
         &self,
         transceiver: Option<rtc::rtp_transceiver::RtpTransceiver>,
     ) {
-        self.inner.update_transceiver(transceiver)
+        self.inner.track_inner.update_transceiver(transceiver)
     }
 
     #[inline]
     pub(crate) fn update_info(&self, info: proto::TrackInfo) {
-        self.inner.update_info(info)
+        self.inner.track_inner.update_info(info)
+    }
+}
+
+impl LocalAudioTrack {
+    pub fn create_audio_track(
+        name: &str,
+        options: AudioCaptureOptions,
+        source: NativeAudioSource,
+    ) -> LocalAudioTrack {
+        let rtc_track = LkRuntime::instance()
+            .pc_factory
+            .create_audio_track(&rtc::native::create_random_uuid(), source);
+
+        Self::new(name.to_string(), rtc_track, options)
     }
 }
diff --git a/livekit/src/room/track/local_video_track.rs b/livekit/src/room/track/local_video_track.rs
index 51064b3..2a427a3 100644
--- a/livekit/src/room/track/local_video_track.rs
+++ b/livekit/src/room/track/local_video_track.rs
@@ -39,6 +39,7 @@ impl LocalVideoTrack {
         }
     }
 
+    #[inline]
     pub fn capture_options(&self) -> VideoCaptureOptions {
         self.inner.capture_options.lock().clone()
     }
diff --git a/livekit/src/rtc_engine/rtc_session.rs b/livekit/src/rtc_engine/rtc_session.rs
index 86e40fc..f742fd9 100644
--- a/livekit/src/rtc_engine/rtc_session.rs
+++ b/livekit/src/rtc_engine/rtc_session.rs
@@ -1,5 +1,6 @@
 use super::{rtc_events, EngineError, EngineResult, SimulateScenario};
 use crate::options::TrackPublishOptions;
+use crate::prelude::TrackKind;
 use crate::rtc_engine::lk_runtime::LkRuntime;
 use crate::rtc_engine::peer_transport::PeerTransport;
 use crate::rtc_engine::rtc_events::{RtcEvent, RtcEvents};
@@ -652,38 +653,38 @@ impl SessionInner {
             .peer_connection()
             .add_transceiver(track.rtc_track(), init)?;
 
-        let capabilities = LkRuntime::instance()
-            .pc_factory
-            .get_rtp_sender_capabilities(track.kind().into());
+        if track.kind() == TrackKind::Video {
+            let capabilities = LkRuntime::instance()
+                .pc_factory
+                .get_rtp_sender_capabilities(track.kind().into());
 
-        let mut matched = Vec::new();
-        let mut partial_matched = Vec::new();
-        let mut unmatched = Vec::new();
+            let mut matched = Vec::new();
+            let mut partial_matched = Vec::new();
+            let mut unmatched = Vec::new();
 
-        for codec in capabilities.codecs {
-            let mime_type = codec.mime_type.to_lowercase();
-            if mime_type == "audio/opus" {
-                matched.push(codec);
-            } else if mime_type == format!("video/{}", options.video_codec.as_str()) {
-                if let Some(sdp_fmtp_line) = codec.sdp_fmtp_line.as_ref() {
-                    // for h264 codecs that have sdpFmtpLine available, use only if the
-                    // profile-level-id is 42e01f for cross-browser compatibility
-                    if sdp_fmtp_line.contains("profile-level-id=42e01f") {
-                        matched.push(codec);
-                        continue;
+            for codec in capabilities.codecs {
+                let mime_type = codec.mime_type.to_lowercase();
+                if mime_type == format!("video/{}", options.video_codec.as_str()) {
+                    if let Some(sdp_fmtp_line) = codec.sdp_fmtp_line.as_ref() {
+                        // for h264 codecs that have sdpFmtpLine available, use only if the
+                        // profile-level-id is 42e01f for cross-browser compatibility
+                        if sdp_fmtp_line.contains("profile-level-id=42e01f") {
+                            matched.push(codec);
+                            continue;
+                        }
                     }
+                    partial_matched.push(codec);
+                } else {
+                    unmatched.push(codec);
                 }
-                partial_matched.push(codec);
-            } else {
-                unmatched.push(codec);
             }
+
+            matched.append(&mut partial_matched);
+            matched.append(&mut unmatched);
+
+            transceiver.set_codec_preferences(matched)?;
         }
 
-        matched.append(&mut partial_matched);
-        matched.append(&mut unmatched);
-
-        transceiver.set_codec_preferences(matched)?;
-
         Ok(transceiver)
     }
 
diff --git a/webrtc-sys/build.rs b/webrtc-sys/build.rs
index 9291961..7cae89c 100644
--- a/webrtc-sys/build.rs
+++ b/webrtc-sys/build.rs
@@ -153,6 +153,7 @@ fn main() {
     builder.file("src/video_frame_buffer.cpp");
     builder.file("src/video_encoder_factory.cpp");
     builder.file("src/video_decoder_factory.cpp");
+    builder.file("src/audio_device.cpp");
 
     for include in includes {
         builder.include(include);
diff --git a/webrtc-sys/include/livekit/audio_device.h b/webrtc-sys/include/livekit/audio_device.h
new file mode 100644
index 0000000..135e178
--- /dev/null
+++ b/webrtc-sys/include/livekit/audio_device.h
@@ -0,0 +1,129 @@
+/*
+ * Copyright 2023 LiveKit
+ *
+ * Licensed under the Apache License, Version 2.0 (the “License”);
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an “AS IS” BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#pragma once
+
+#include <atomic>
+
+#include "api/task_queue/task_queue_factory.h"
+#include "modules/audio_device/include/audio_device.h"
+#include "rtc_base/synchronization/mutex.h"
+#include "rtc_base/task_queue.h"
+#include "rtc_base/task_utils/repeating_task.h"
+
+namespace livekit {
+
+class AudioDevice : public webrtc::AudioDeviceModule {
+ public:
+  AudioDevice(webrtc::TaskQueueFactory* task_queue_factory);
+  ~AudioDevice() override;
+
+  int32_t ActiveAudioLayer(AudioLayer* audioLayer) const override;
+  int32_t RegisterAudioCallback(webrtc::AudioTransport* transport) override;
+
+  int32_t Init() override;
+  int32_t Terminate() override;
+  bool Initialized() const override;
+
+  int16_t PlayoutDevices() override;
+  int16_t RecordingDevices() override;
+  int32_t PlayoutDeviceName(uint16_t index,
+                            char name[webrtc::kAdmMaxDeviceNameSize],
+                            char guid[webrtc::kAdmMaxGuidSize]) override;
+
+  int32_t RecordingDeviceName(uint16_t index,
+                              char name[webrtc::kAdmMaxDeviceNameSize],
+                              char guid[webrtc::kAdmMaxGuidSize]) override;
+
+  int32_t SetPlayoutDevice(uint16_t index) override;
+  int32_t SetPlayoutDevice(WindowsDeviceType device) override;
+  int32_t SetRecordingDevice(uint16_t index) override;
+  int32_t SetRecordingDevice(WindowsDeviceType device) override;
+
+  int32_t PlayoutIsAvailable(bool* available) override;
+  int32_t InitPlayout() override;
+  bool PlayoutIsInitialized() const override;
+  int32_t RecordingIsAvailable(bool* available) override;
+  int32_t InitRecording() override;
+  bool RecordingIsInitialized() const override;
+
+  int32_t StartPlayout() override;
+  int32_t StopPlayout() override;
+  bool Playing() const override;
+  int32_t StartRecording() override;
+  int32_t StopRecording() override;
+  bool Recording() const override;
+
+  int32_t InitSpeaker() override;
+  bool SpeakerIsInitialized() const override;
+  int32_t InitMicrophone() override;
+  bool MicrophoneIsInitialized() const override;
+
+  int32_t SpeakerVolumeIsAvailable(bool* available) override;
+  int32_t SetSpeakerVolume(uint32_t volume) override;
+  int32_t SpeakerVolume(uint32_t* volume) const override;
+  int32_t MaxSpeakerVolume(uint32_t* maxVolume) const override;
+  int32_t MinSpeakerVolume(uint32_t* minVolume) const override;
+
+  int32_t MicrophoneVolumeIsAvailable(bool* available) override;
+  int32_t SetMicrophoneVolume(uint32_t volume) override;
+  int32_t MicrophoneVolume(uint32_t* volume) const override;
+  int32_t MaxMicrophoneVolume(uint32_t* maxVolume) const override;
+  int32_t MinMicrophoneVolume(uint32_t* minVolume) const override;
+
+  int32_t SpeakerMuteIsAvailable(bool* available) override;
+  int32_t SetSpeakerMute(bool enable) override;
+  int32_t SpeakerMute(bool* enabled) const override;
+
+  int32_t MicrophoneMuteIsAvailable(bool* available) override;
+  int32_t SetMicrophoneMute(bool enable) override;
+  int32_t MicrophoneMute(bool* enabled) const override;
+
+  int32_t StereoPlayoutIsAvailable(bool* available) const override;
+  int32_t SetStereoPlayout(bool enable) override;
+  int32_t StereoPlayout(bool* enabled) const override;
+  int32_t StereoRecordingIsAvailable(bool* available) const override;
+  int32_t SetStereoRecording(bool enable) override;
+  int32_t StereoRecording(bool* enabled) const override;
+
+  int32_t PlayoutDelay(uint16_t* delayMS) const override;
+
+  bool BuiltInAECIsAvailable() const override;
+  bool BuiltInAGCIsAvailable() const override;
+  bool BuiltInNSIsAvailable() const override;
+
+  int32_t EnableBuiltInAEC(bool enable) override;
+  int32_t EnableBuiltInAGC(bool enable) override;
+  int32_t EnableBuiltInNS(bool enable) override;
+
+#if defined(WEBRTC_IOS)
+  int GetPlayoutAudioParameters(AudioParameters* params) const override;
+  int GetRecordAudioParameters(AudioParameters* params) const override;
+#endif  // WEBRTC_IOS
+
+  int32_t SetAudioDeviceSink(webrtc::AudioDeviceSink* sink) const override;
+
+ private:
+  mutable webrtc::Mutex mutex_;
+  webrtc::TaskQueueFactory* task_queue_factory_;
+  std::unique_ptr<rtc::TaskQueue> audio_queue_;
+  webrtc::RepeatingTaskHandle audio_task_;
+  std::vector<int16_t> data_;
+  webrtc::AudioTransport* audio_transport_;
+  std::atomic<bool> playing_{false};
+  std::atomic<bool> initialized_{false};
+};
+}  // namespace livekit
diff --git a/webrtc-sys/include/livekit/media_stream.h b/webrtc-sys/include/livekit/media_stream.h
index 8960791..6d94bfc 100644
--- a/webrtc-sys/include/livekit/media_stream.h
+++ b/webrtc-sys/include/livekit/media_stream.h
@@ -20,9 +20,12 @@
 
 #include "api/media_stream_interface.h"
 #include "api/video/video_frame.h"
+#include "common_audio/resampler/include/push_resampler.h"
+#include "common_audio/ring_buffer.h"
 #include "livekit/helper.h"
 #include "livekit/video_frame.h"
 #include "media/base/adapted_video_track_source.h"
+#include "pc/local_audio_source.h"
 #include "rtc_base/synchronization/mutex.h"
 #include "rtc_base/timestamp_aligner.h"
 #include "rust/cxx.h"
@@ -34,6 +37,8 @@ class MediaStreamTrack;
 class VideoTrack;
 class AudioTrack;
 class NativeVideoFrameSink;
+class NativeAudioSink;
+class AudioTrackSource;
 class AdaptedVideoTrackSource;
 }  // namespace livekit
 #include "webrtc-sys/src/media_stream.rs.h"
@@ -86,8 +91,73 @@ class MediaStreamTrack {
 class AudioTrack : public MediaStreamTrack {
  public:
   explicit AudioTrack(rtc::scoped_refptr<webrtc::AudioTrackInterface> track);
+
+  void add_sink(NativeAudioSink& sink) const;
+  void remove_sink(NativeAudioSink& sink) const;
+
+ private:
+  webrtc::AudioTrackInterface* track() const {
+    return static_cast<webrtc::AudioTrackInterface*>(track_.get());
+  }
 };
 
+class NativeAudioSink : public webrtc::AudioTrackSinkInterface {
+ public:
+  explicit NativeAudioSink(rust::Box<AudioSinkWrapper> observer);
+  void OnData(const void* audio_data,
+              int bits_per_sample,
+              int sample_rate,
+              size_t number_of_channels,
+              size_t number_of_frames) override;
+
+ private:
+  rust::Box<AudioSinkWrapper> observer_;
+};
+
+std::unique_ptr<NativeAudioSink> new_native_audio_sink(
+    rust::Box<AudioSinkWrapper> observer);
+
+class NativeAudioTrackSource : public webrtc::LocalAudioSource {
+ public:
+  NativeAudioTrackSource();
+
+  SourceState state() const override;
+  bool remote() const override;
+
+  const cricket::AudioOptions options() const override;
+
+  void AddSink(webrtc::AudioTrackSinkInterface* sink) override;
+  void RemoveSink(webrtc::AudioTrackSinkInterface* sink) override;
+
+  // AudioFrame should always contain 10 ms worth of data (see index.md of acm)
+  void on_captured_frame(const int16_t* audio_data,
+                         int sample_rate,
+                         size_t number_of_channels,
+                         size_t number_of_frames);
+
+ private:
+  webrtc::Mutex mutex_;
+  std::vector<webrtc::AudioTrackSinkInterface*> sinks_;
+  cricket::AudioOptions options_{};
+};
+
+class AudioTrackSource {
+ public:
+  AudioTrackSource(rtc::scoped_refptr<NativeAudioTrackSource> source);
+
+  void on_captured_frame(const int16_t* audio_data,
+                         int sample_rate,
+                         size_t number_of_channels,
+                         size_t number_of_frames) const;
+
+  rtc::scoped_refptr<NativeAudioTrackSource> get() const;
+
+ private:
+  rtc::scoped_refptr<NativeAudioTrackSource> source_;
+};
+
+std::shared_ptr<AudioTrackSource> new_audio_track_source();
+
 class VideoTrack : public MediaStreamTrack {
  public:
   explicit VideoTrack(rtc::scoped_refptr<webrtc::VideoTrackInterface> track);
@@ -131,7 +201,7 @@ class NativeVideoTrackSource : public rtc::AdaptedVideoTrackSource {
 
   bool is_screencast() const override;
   absl::optional<bool> needs_denoising() const override;
-  webrtc::MediaSourceInterface::SourceState state() const override;
+  SourceState state() const override;
   bool remote() const override;
 
   bool on_captured_frame(const webrtc::VideoFrame& frame);
diff --git a/webrtc-sys/include/livekit/peer_connection_factory.h b/webrtc-sys/include/livekit/peer_connection_factory.h
index 5e7fcba..884bcef 100644
--- a/webrtc-sys/include/livekit/peer_connection_factory.h
+++ b/webrtc-sys/include/livekit/peer_connection_factory.h
@@ -46,6 +46,10 @@ class PeerConnectionFactory {
       rust::String label,
       std::shared_ptr<AdaptedVideoTrackSource> source) const;
 
+  std::shared_ptr<AudioTrack> create_audio_track(
+      rust::String label,
+      std::shared_ptr<AudioTrackSource> source) const;
+
   RtpCapabilities get_rtp_sender_capabilities(MediaType type) const;
 
   RtpCapabilities get_rtp_receiver_capabilities(MediaType type) const;
diff --git a/webrtc-sys/src/audio_device.cpp b/webrtc-sys/src/audio_device.cpp
new file mode 100644
index 0000000..7bc35c5
--- /dev/null
+++ b/webrtc-sys/src/audio_device.cpp
@@ -0,0 +1,323 @@
+/*
+ * Copyright 2023 LiveKit
+ *
+ * Licensed under the Apache License, Version 2.0 (the “License”);
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an “AS IS” BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "livekit/audio_device.h"
+
+const int kBitsPerSample = 16;
+const int kSampleRate = 48000;
+const int kChannels = 2;
+const int kSamplesPer10Ms = kSampleRate / 100;
+
+namespace livekit {
+
+AudioDevice::AudioDevice(webrtc::TaskQueueFactory* task_queue_factory)
+    : task_queue_factory_(task_queue_factory),
+      data_(kSamplesPer10Ms * kChannels) {}
+
+AudioDevice::~AudioDevice() {
+  Terminate();
+}
+
+int32_t AudioDevice::ActiveAudioLayer(AudioLayer* audioLayer) const {
+  *audioLayer = AudioLayer::kDummyAudio;
+  return 0;
+}
+
+int32_t AudioDevice::RegisterAudioCallback(webrtc::AudioTransport* transport) {
+  webrtc::MutexLock lock(&mutex_);
+  audio_transport_ = transport;
+  return 0;
+}
+
+int32_t AudioDevice::Init() {
+  audio_queue_ =
+      std::make_unique<rtc::TaskQueue>(task_queue_factory_->CreateTaskQueue(
+          "AudioDevice", webrtc::TaskQueueFactory::Priority::NORMAL));
+
+  audio_task_ =
+      webrtc::RepeatingTaskHandle::Start(audio_queue_->Get(), [this]() {
+        webrtc::MutexLock lock(&mutex_);
+
+        if (playing_) {
+          int64_t elapsed_time_ms = -1;
+          int64_t ntp_time_ms = -1;
+          void* data = data_.data();
+
+          // Request the AudioData, otherwise WebRTC will ignore the packets.
+          // 10ms of audio data.
+          audio_transport_->PullRenderData(kBitsPerSample, kSampleRate,
+                                           kChannels, kSamplesPer10Ms, data,
+                                           &elapsed_time_ms, &ntp_time_ms);
+        }
+
+        return webrtc::TimeDelta::Millis(10);
+      });
+
+  initialized_ = true;
+  return 0;
+}
+
+int32_t AudioDevice::Terminate() {
+  if (!initialized_)
+    return 0;
+
+  initialized_ = false;
+
+  audio_queue_->PostTask([this] { audio_task_.Stop(); });
+
+  StopRecording();
+  StopPlayout();
+  return 0;
+}
+
+bool AudioDevice::Initialized() const {
+  return initialized_;
+}
+
+int16_t AudioDevice::PlayoutDevices() {
+  return 0;
+}
+
+int16_t AudioDevice::RecordingDevices() {
+  return 0;
+}
+
+int32_t AudioDevice::PlayoutDeviceName(uint16_t index,
+                                       char name[webrtc::kAdmMaxDeviceNameSize],
+                                       char guid[webrtc::kAdmMaxGuidSize]) {
+  return 0;
+}
+
+int32_t AudioDevice::RecordingDeviceName(
+    uint16_t index,
+    char name[webrtc::kAdmMaxDeviceNameSize],
+    char guid[webrtc::kAdmMaxGuidSize]) {
+  return 0;
+}
+
+int32_t AudioDevice::SetPlayoutDevice(uint16_t index) {
+  return 0;
+}
+
+int32_t AudioDevice::SetPlayoutDevice(WindowsDeviceType device) {
+  return 0;
+}
+
+int32_t AudioDevice::SetRecordingDevice(uint16_t index) {
+  return 0;
+}
+
+int32_t AudioDevice::SetRecordingDevice(WindowsDeviceType device) {
+  return 0;
+}
+
+int32_t AudioDevice::PlayoutIsAvailable(bool* available) {
+  return 0;
+}
+
+int32_t AudioDevice::InitPlayout() {
+  return 0;
+}
+
+bool AudioDevice::PlayoutIsInitialized() const {
+  return false;
+}
+
+int32_t AudioDevice::RecordingIsAvailable(bool* available) {
+  return 0;
+}
+
+int32_t AudioDevice::InitRecording() {
+  return 0;
+}
+
+bool AudioDevice::RecordingIsInitialized() const {
+  return false;
+}
+
+int32_t AudioDevice::StartPlayout() {
+  playing_ = true;
+  return 0;
+}
+
+int32_t AudioDevice::StopPlayout() {
+  playing_ = false;
+  return 0;
+}
+
+bool AudioDevice::Playing() const {
+  return false;
+}
+
+int32_t AudioDevice::StartRecording() {
+  return 0;
+}
+
+int32_t AudioDevice::StopRecording() {
+  return 0;
+}
+
+bool AudioDevice::Recording() const {
+  return false;
+}
+
+int32_t AudioDevice::InitSpeaker() {
+  return 0;
+}
+
+bool AudioDevice::SpeakerIsInitialized() const {
+  return false;
+}
+
+int32_t AudioDevice::InitMicrophone() {
+  return 0;
+}
+
+bool AudioDevice::MicrophoneIsInitialized() const {
+  return false;
+}
+
+int32_t AudioDevice::SpeakerVolumeIsAvailable(bool* available) {
+  return 0;
+}
+
+int32_t AudioDevice::SetSpeakerVolume(uint32_t volume) {
+  return 0;
+}
+
+int32_t AudioDevice::SpeakerVolume(uint32_t* volume) const {
+  return 0;
+}
+
+int32_t AudioDevice::MaxSpeakerVolume(uint32_t* maxVolume) const {
+  return 0;
+}
+
+int32_t AudioDevice::MinSpeakerVolume(uint32_t* minVolume) const {
+  return 0;
+}
+
+int32_t AudioDevice::MicrophoneVolumeIsAvailable(bool* available) {
+  return 0;
+}
+
+int32_t AudioDevice::SetMicrophoneVolume(uint32_t volume) {
+  return 0;
+}
+
+int32_t AudioDevice::MicrophoneVolume(uint32_t* volume) const {
+  return 0;
+}
+
+int32_t AudioDevice::MaxMicrophoneVolume(uint32_t* maxVolume) const {
+  return 0;
+}
+
+int32_t AudioDevice::MinMicrophoneVolume(uint32_t* minVolume) const {
+  return 0;
+}
+
+int32_t AudioDevice::SpeakerMuteIsAvailable(bool* available) {
+  return 0;
+}
+
+int32_t AudioDevice::SetSpeakerMute(bool enable) {
+  return 0;
+}
+
+int32_t AudioDevice::SpeakerMute(bool* enabled) const {
+  return 0;
+}
+
+int32_t AudioDevice::MicrophoneMuteIsAvailable(bool* available) {
+  return 0;
+}
+
+int32_t AudioDevice::SetMicrophoneMute(bool enable) {
+  return 0;
+}
+
+int32_t AudioDevice::MicrophoneMute(bool* enabled) const {
+  return 0;
+}
+
+int32_t AudioDevice::StereoPlayoutIsAvailable(bool* available) const {
+  return 0;
+}
+
+int32_t AudioDevice::SetStereoPlayout(bool enable) {
+  return 0;
+}
+
+int32_t AudioDevice::StereoPlayout(bool* enabled) const {
+  return 0;
+}
+
+int32_t AudioDevice::StereoRecordingIsAvailable(bool* available) const {
+  return 0;
+}
+
+int32_t AudioDevice::SetStereoRecording(bool enable) {
+  return 0;
+}
+
+int32_t AudioDevice::StereoRecording(bool* enabled) const {
+  return 0;
+}
+
+int32_t AudioDevice::PlayoutDelay(uint16_t* delayMS) const {
+  return 0;
+}
+
+bool AudioDevice::BuiltInAECIsAvailable() const {
+  return false;
+}
+
+bool AudioDevice::BuiltInAGCIsAvailable() const {
+  return false;
+}
+
+bool AudioDevice::BuiltInNSIsAvailable() const {
+  return false;
+}
+
+int32_t AudioDevice::EnableBuiltInAEC(bool enable) {
+  return 0;
+}
+
+int32_t AudioDevice::EnableBuiltInAGC(bool enable) {
+  return 0;
+}
+
+int32_t AudioDevice::EnableBuiltInNS(bool enable) {
+  return 0;
+}
+
+#if defined(WEBRTC_IOS)
+int AudioDevice::GetPlayoutAudioParameters(AudioParameters* params) const {
+  return 0;
+}
+
+int AudioDevice::GetRecordAudioParameters(AudioParameters* params) const {
+  return 0;
+}
+#endif  // WEBRTC_IOS
+
+int32_t AudioDevice::SetAudioDeviceSink(webrtc::AudioDeviceSink* sink) const {
+  return 0;
+}
+
+}  // namespace livekit
diff --git a/webrtc-sys/src/media_stream.cpp b/webrtc-sys/src/media_stream.cpp
index a0839a0..cbd8c96 100644
--- a/webrtc-sys/src/media_stream.cpp
+++ b/webrtc-sys/src/media_stream.cpp
@@ -23,6 +23,8 @@
 #include "api/media_stream_interface.h"
 #include "api/video/video_frame.h"
 #include "api/video/video_rotation.h"
+#include "audio/remix_resample.h"
+#include "common_audio/include/audio_util.h"
 #include "rtc_base/logging.h"
 #include "rtc_base/ref_counted_object.h"
 #include "rtc_base/time_utils.h"
@@ -129,6 +131,93 @@ TrackState MediaStreamTrack::state() const {
 AudioTrack::AudioTrack(rtc::scoped_refptr<webrtc::AudioTrackInterface> track)
     : MediaStreamTrack(std::move(track)) {}
 
+void AudioTrack::add_sink(NativeAudioSink& sink) const {
+  track()->AddSink(&sink);
+}
+
+void AudioTrack::remove_sink(NativeAudioSink& sink) const {
+  track()->RemoveSink(&sink);
+}
+
+NativeAudioSink::NativeAudioSink(rust::Box<AudioSinkWrapper> observer)
+    : observer_(std::move(observer)) {}
+
+void NativeAudioSink::OnData(const void* audio_data,
+                             int bits_per_sample,
+                             int sample_rate,
+                             size_t number_of_channels,
+                             size_t number_of_frames) {
+  RTC_CHECK_EQ(16, bits_per_sample);
+
+  observer_->on_data(static_cast<const int16_t*>(audio_data), sample_rate,
+                     number_of_channels, number_of_frames);
+}
+
+std::unique_ptr<NativeAudioSink> new_native_audio_sink(
+    rust::Box<AudioSinkWrapper> observer) {
+  return std::make_unique<NativeAudioSink>(std::move(observer));
+}
+
+NativeAudioTrackSource::NativeAudioTrackSource() {
+  options_.echo_cancellation = false;
+  options_.auto_gain_control = false;
+  options_.noise_suppression = false;
+}
+
+webrtc::MediaSourceInterface::SourceState NativeAudioTrackSource::state()
+    const {
+  return webrtc::MediaSourceInterface::SourceState::kLive;
+}
+
+bool NativeAudioTrackSource::remote() const {
+  return false;
+}
+
+const cricket::AudioOptions NativeAudioTrackSource::options() const {
+  return options_;
+}
+
+void NativeAudioTrackSource::AddSink(webrtc::AudioTrackSinkInterface* sink) {
+  webrtc::MutexLock lock(&mutex_);
+  sinks_.push_back(sink);
+}
+
+void NativeAudioTrackSource::RemoveSink(webrtc::AudioTrackSinkInterface* sink) {
+  webrtc::MutexLock lock(&mutex_);
+  sinks_.erase(std::remove(sinks_.begin(), sinks_.end(), sink), sinks_.end());
+}
+
+void NativeAudioTrackSource::on_captured_frame(const int16_t* data,
+                                               int sample_rate,
+                                               size_t number_of_channels,
+                                               size_t number_of_frames) {
+  webrtc::MutexLock lock(&mutex_);
+  for (auto sink : sinks_) {
+    sink->OnData(data, 16, sample_rate, number_of_channels, number_of_frames);
+  }
+}
+
+AudioTrackSource::AudioTrackSource(
+    rtc::scoped_refptr<NativeAudioTrackSource> source)
+    : source_(std::move(source)) {}
+
+void AudioTrackSource::on_captured_frame(const int16_t* audio_data,
+                                         int sample_rate,
+                                         size_t number_of_channels,
+                                         size_t number_of_frames) const {
+  source_->on_captured_frame(audio_data, sample_rate, number_of_channels,
+                             number_of_frames);
+}
+
+rtc::scoped_refptr<NativeAudioTrackSource> AudioTrackSource::get() const {
+  return source_;
+}
+
+std::shared_ptr<AudioTrackSource> new_audio_track_source() {
+  return std::make_shared<AudioTrackSource>(
+      rtc::make_ref_counted<NativeAudioTrackSource>());
+}
+
 VideoTrack::VideoTrack(rtc::scoped_refptr<webrtc::VideoTrackInterface> track)
     : MediaStreamTrack(std::move(track)) {}
 
diff --git a/webrtc-sys/src/media_stream.rs b/webrtc-sys/src/media_stream.rs
index 2afbd33..77fcabf 100644
--- a/webrtc-sys/src/media_stream.rs
+++ b/webrtc-sys/src/media_stream.rs
@@ -40,11 +40,13 @@ pub mod ffi {
     unsafe extern "C++" {
         include!("livekit/media_stream.h");
 
+        type NativeAudioSink;
         type NativeVideoFrameSink;
         type MediaStreamTrack;
         type MediaStream;
         type AudioTrack;
         type VideoTrack;
+        type AudioTrackSource;
         type AdaptedVideoTrackSource;
 
         fn id(self: &MediaStream) -> String;
@@ -61,6 +63,21 @@ pub mod ffi {
         fn set_enabled(self: &MediaStreamTrack, enable: bool) -> bool;
         fn state(self: &MediaStreamTrack) -> TrackState;
 
+        unsafe fn add_sink(self: &AudioTrack, sink: Pin<&mut NativeAudioSink>);
+        unsafe fn remove_sink(self: &AudioTrack, sink: Pin<&mut NativeAudioSink>);
+
+        fn new_native_audio_sink(observer: Box<AudioSinkWrapper>) -> UniquePtr<NativeAudioSink>;
+
+        unsafe fn on_captured_frame(
+            self: &AudioTrackSource,
+            data: *const i16,
+            sample_rate: i32,
+            nb_channels: usize,
+            nb_frames: usize,
+        );
+
+        fn new_audio_track_source() -> SharedPtr<AudioTrackSource>;
+
         unsafe fn add_sink(self: &VideoTrack, sink: Pin<&mut NativeVideoFrameSink>);
         unsafe fn remove_sink(self: &VideoTrack, sink: Pin<&mut NativeVideoFrameSink>);
 
@@ -90,8 +107,17 @@ pub mod ffi {
     }
 
     extern "Rust" {
+        type AudioSinkWrapper;
         type VideoFrameSinkWrapper;
 
+        unsafe fn on_data(
+            self: &AudioSinkWrapper,
+            data: *const i16,
+            sample_rate: i32,
+            nb_channels: usize,
+            nb_frames: usize,
+        );
+
         fn on_frame(self: &VideoFrameSinkWrapper, frame: UniquePtr<VideoFrame>);
         fn on_discarded_frame(self: &VideoFrameSinkWrapper);
         fn on_constraints_changed(
@@ -106,8 +132,33 @@ impl_thread_safety!(ffi::MediaStream, Send + Sync);
 impl_thread_safety!(ffi::AudioTrack, Send + Sync);
 impl_thread_safety!(ffi::VideoTrack, Send + Sync);
 impl_thread_safety!(ffi::NativeVideoFrameSink, Send + Sync);
+impl_thread_safety!(ffi::NativeAudioSink, Send + Sync);
+impl_thread_safety!(ffi::AudioTrackSource, Send + Sync);
 impl_thread_safety!(ffi::AdaptedVideoTrackSource, Send + Sync);
 
+pub trait AudioSink: Send {
+    fn on_data(&self, data: &[i16], sample_rate: i32, nb_channels: usize, nb_frames: usize);
+}
+
+pub struct AudioSinkWrapper {
+    observer: *mut dyn AudioSink,
+}
+
+impl AudioSinkWrapper {
+    /// # Safety
+    /// AudioSink must lives as long as AudioSinkWrapper does
+    pub unsafe fn new(observer: *mut dyn AudioSink) -> Self {
+        Self { observer }
+    }
+
+    fn on_data(&self, data: *const i16, sample_rate: i32, nb_channels: usize, nb_frames: usize) {
+        unsafe {
+            let data = std::slice::from_raw_parts(data, nb_channels * nb_frames);
+            (*self.observer).on_data(data, sample_rate, nb_channels, nb_frames);
+        }
+    }
+}
+
 pub trait VideoFrameSink: Send {
     fn on_frame(&self, frame: UniquePtr<VideoFrame>);
     fn on_discarded_frame(&self);
diff --git a/webrtc-sys/src/peer_connection_factory.cpp b/webrtc-sys/src/peer_connection_factory.cpp
index 6a2281c..9e863bc 100644
--- a/webrtc-sys/src/peer_connection_factory.cpp
+++ b/webrtc-sys/src/peer_connection_factory.cpp
@@ -25,11 +25,14 @@
 #include "api/task_queue/default_task_queue_factory.h"
 #include "api/video_codecs/builtin_video_decoder_factory.h"
 #include "api/video_codecs/builtin_video_encoder_factory.h"
+#include "livekit/audio_device.h"
 #include "livekit/rtc_error.h"
 #include "livekit/rtp_parameters.h"
 #include "livekit/video_decoder_factory.h"
 #include "livekit/video_encoder_factory.h"
 #include "media/engine/webrtc_media_engine.h"
+#include "rtc_base/location.h"
+#include "rtc_base/thread.h"
 
 namespace livekit {
 
@@ -51,6 +54,14 @@ PeerConnectionFactory::PeerConnectionFactory(
 
   cricket::MediaEngineDependencies media_deps;
   media_deps.task_queue_factory = dependencies.task_queue_factory.get();
+
+  media_deps.adm = rtc_runtime_->worker_thread()
+                       ->Invoke<rtc::scoped_refptr<livekit::AudioDevice>>(
+                           RTC_FROM_HERE, [&] {
+                             return rtc::make_ref_counted<livekit::AudioDevice>(
+                                 media_deps.task_queue_factory);
+                           });
+
   media_deps.video_encoder_factory =
       std::move(std::make_unique<livekit::VideoEncoderFactory>());
   media_deps.video_decoder_factory =
@@ -96,6 +107,13 @@ std::shared_ptr<VideoTrack> PeerConnectionFactory::create_video_track(
       peer_factory_->CreateVideoTrack(label.c_str(), source->get().get()));
 }
 
+std::shared_ptr<AudioTrack> PeerConnectionFactory::create_audio_track(
+    rust::String label,
+    std::shared_ptr<AudioTrackSource> source) const {
+  return std::make_shared<AudioTrack>(
+      peer_factory_->CreateAudioTrack(label.c_str(), source->get().get()));
+}
+
 RtpCapabilities PeerConnectionFactory::get_rtp_sender_capabilities(
     MediaType type) const {
   return to_rust_rtp_capabilities(peer_factory_->GetRtpSenderCapabilities(
diff --git a/webrtc-sys/src/peer_connection_factory.rs b/webrtc-sys/src/peer_connection_factory.rs
index a099c64..3561fde 100644
--- a/webrtc-sys/src/peer_connection_factory.rs
+++ b/webrtc-sys/src/peer_connection_factory.rs
@@ -37,7 +37,9 @@ pub mod ffi {
         include!("livekit/webrtc.h");
         include!("livekit/rtp_parameters.h");
 
+        type AudioTrackSource = crate::media_stream::ffi::AudioTrackSource;
         type AdaptedVideoTrackSource = crate::media_stream::ffi::AdaptedVideoTrackSource;
+        type AudioTrack = crate::media_stream::ffi::AudioTrack;
         type VideoTrack = crate::media_stream::ffi::VideoTrack;
         type RtpCapabilities = crate::rtp_parameters::ffi::RtpCapabilities;
         type MediaType = crate::webrtc::ffi::MediaType;
@@ -72,6 +74,12 @@ pub mod ffi {
             source: SharedPtr<AdaptedVideoTrackSource>,
         ) -> SharedPtr<VideoTrack>;
 
+        fn create_audio_track(
+            self: &PeerConnectionFactory,
+            label: String,
+            source: SharedPtr<AudioTrackSource>,
+        ) -> SharedPtr<AudioTrack>;
+
         fn get_rtp_sender_capabilities(
             self: &PeerConnectionFactory,
             kind: MediaType,