Skip to content

relay.md

Latest commit

 

History

History
58 lines (44 loc) · 2.13 KB

relay.md

File metadata and controls

58 lines (44 loc) · 2.13 KB
Raw

Publish a named AudioBus through Relay

Use RelaySession when a Session stem must reach a browser or another remote receiver. The Python control client creates credentials and invitations; the shared Rust Connector handles WebRTC publication.

Connect to services you operate

import pocketstation.aio as pks

remote = await pks.RelaySession.create(
    control_plane_url="https://control.example.com",
    relay_url="https://relay.example.com",
    required_buses=("application",),
)
live = pks.capture(application="Spotify", stream_audio=False)
live.application_stem.publish(remote.publisher(live.session), "application")

async with remote, live:
    invitation = await remote.wait_for_publisher_and_invitation(
        bus_id="application",
        timeout_seconds=30,
    )
    print(invitation.join_code, invitation.join_url)
    await remote.wait_for_receiver(timeout_seconds=30)

Declare every bus before starting the Session. Create an invitation only after publisher readiness succeeds, and delete the remote Session during shutdown.

Use the shared demo service for a quick test

examples/stream_any_app_audio.py uses the small rate-limited demo deployment through pocketstation_demo. The deployment may reject a session when its capacity is in use and is not a hosted production service.

Set these variables to run the same example against services you operate:

export POCKETSTATION_CONTROL_URL="https://control.example.com"
export POCKETSTATION_RELAY_URL="https://relay.example.com"
python examples/stream_any_app_audio.py

Do not put control-plane secrets, signing keys, or shared internal credentials in application code. Applications receive scoped session credentials from the control plane.

Know what readiness proves

Publisher readiness confirms that Relay accepted the declared publication. Receiver readiness confirms an active subscription. Browser WebRTC statistics can report received and jitter-buffered samples.

Those observations do not prove which sample a loudspeaker played. End-to-end audible cancellation requires a receiver capability that clears playout and acknowledges the last rendered sample.