< SDK Flutter headless />
UTD Stream SDK
Sessions audio/vidéo LiveKit headless pour Flutter — l'UI, c'est vous qui la construisez.
final client = UTDStreamClient(
appId: '1234567890',
appKey: 'a1b2c3d4e5f6a7b8c9d0e1f2a3b4c5d6',
);
// Audio room
final room = await client.joinAudioRoom(
identity: 'u1', roomName: 'lobby', asHost: true, seatCount: 8,
);
await room.takeSeat(0);
// Live stream
final live = await client.joinLiveStream(
identity: 'host1', roomName: 'show', asHost: true,
);
await live.goLive();
// 1:1 call
await client.authenticate(identity: 'u1');
final call = client.callSession(UTDStreamType.videoCall);
await call.start(calleeIdentity: 'u2');4
Types de sessions
Headless
Zéro UI
7
Plateformes
LiveKit
Moteur
< utd_stream_sdk />
Fonctionnalités clés
Un SDK headless pour créer des applications audio/vidéo sous Flutter avec LiveKit comme moteur. Il expose quatre types de sessions (appel audio, appel vidéo, live stream, salon audio) en tant que sessions purement logiques, exposant leur état via des ValueListenables et un flux d'événements brut. Toute l'UI vous appartient ; le SDK gère les sessions, les permissions faisant autorité côté serveur et le contrôle des médias — sans aucun secret embarqué dans l'application.
Quatre types de sessions : appel audio, appel vidéo, live stream, salon audio
Headless par conception — pure logique et état, zéro UI
Intégration LiveKit avec streaming adaptatif et dynacast activé par défaut
État réactif via des ValueListenables, plus un flux RoomEvent brut
Permissions faisant autorité côté serveur (promotion sur siège/scène pilotée par le moteur)
Gestion de scène en live stream avec un nombre illimité de co-publishers
Modèle intervenant/auditeur basé sur les sièges pour les salons audio
Machine à états du cycle de vie des appels 1:1
Façade REST typée, sans aucun secret dans l'application
< utd_stream_sdk />
Commencer
Installer
flutter pub add utd_stream_sdkVoir sur pub.dev
Sessions audio/vidéo LiveKit headless pour Flutter — l'UI, c'est vous qui la construisez.
< utd_stream_sdk />
Référence de l'API
Client & entry point
Top-level entry point that holds credentials and produces logic-only sessions for each product type.
UTDStreamClientconstructorfactory UTDStreamClient({required String appId, required String appKey, String baseUrl, String tokenBaseUrl})Builds a headless client from publishable credentials (appId + appKey). One instance per signed-in user; shares a single HTTP layer across all sessions.
Paramètres
appIdStringrequisPublishable app id for the project.
appKeyStringrequisPublishable app key paired with appId.
baseUrlStringPar défaut = UTDApiClient.defaultBaseUrlEngine REST base URL.
tokenBaseUrlStringPar défaut = UTDApiClient.defaultTokenBaseUrlToken-minting service base URL.
identitygetterString? get identityThe identity of the currently authenticated / joined user, or null if none.
Retourne: String? — Current user identity or null.
authenticatemethodasyncFuture<String> authenticate({required String identity, String? name, String? deviceId})Mints a room-less per-user bearer and attaches it for privileged calls. Required before starting/accepting 1:1 calls. Returns the user_token.
Paramètres
identityStringrequisUnique user identity to mint the bearer for.
nameString?Optional display name.
deviceIdString?Optional device identifier.
Retourne: Future<String> — The minted user_token bearer.
callSessionmethodUTDCallSession callSession(UTDStreamType type)Creates a 1:1 call session (audio or video). Call authenticate() first. Asserts the type is a call type.
Paramètres
typeUTDStreamTyperequisMust be audioCall or videoCall.
Retourne: UTDCallSession — A new call session driven by start()/accept().
joinLiveStreammethodasyncFuture<UTDLiveStreamSession> joinLiveStream({required String identity, required String roomName, bool asHost, String? name, String? deviceId})Mints a room token, connects the media room, and returns a live stream session. Host publishes via goLive(); viewers subscribe and may requestStage().
Paramètres
identityStringrequisJoining user identity.
roomNameStringrequisRoom to join (un-prefixed name).
asHostboolPar défaut = falseJoin as host (role=host, owns the room) vs audience.
nameString?Optional display name.
deviceIdString?Optional device identifier.
Retourne: Future<UTDLiveStreamSession> — A connected live stream session.
joinAudioRoommethodasyncFuture<UTDAudioRoomSession> joinAudioRoom({required String identity, required String roomName, bool asHost, String? name, String? deviceId, int? seatCount, String? seatMode})Mints a room token, connects, and returns a seat-based audio room session. Host may seed the seat grid via seatCount/seatMode.
Paramètres
identityStringrequisJoining user identity.
roomNameStringrequisRoom to join (un-prefixed name).
asHostboolPar défaut = falseJoin as host vs audience; only hosts may seed seats.
nameString?Optional display name.
deviceIdString?Optional device identifier.
seatCountint?Host only: number of seats to seed the grid with.
seatModeString?Host only: seat mode (e.g. free/request).
Retourne: Future<UTDAudioRoomSession> — A connected audio room session.
disposemethodvoid dispose()Releases the underlying HTTP pools. Sessions must be disposed separately.
Retourne: void — Nothing.
Session base & headless state
Shared base class re-exposing the LiveKit room state as ValueListenables and a raw event stream.
UTDSessionclassabstract class UTDSessionAbstract base for every session type. Owns the UTDRoom and re-exposes its headless state; concrete sessions add per-type control (call lifecycle, stage, seats).
typepropertyfinal UTDStreamType typeThe product type of this session (audioCall, videoCall, liveStream, audioRoom).
Retourne: UTDStreamType — The session product type.
roompropertyfinal UTDRoom roomThe headless UTDRoom wrapper owning the LiveKit connection lifecycle.
Retourne: UTDRoom — The underlying room wrapper.
connectedgetterValueListenable<bool> get connectedListenable that reflects whether the media room is currently connected.
Retourne: ValueListenable<bool> — Connection state.
canPublishgetterValueListenable<bool> get canPublishListenable for whether the server has granted this participant publish capability (seat/stage authoritative).
Retourne: ValueListenable<bool> — Publish capability state.
remoteParticipantsgetterValueListenable<List<RemoteParticipant>> get remoteParticipantsListenable list of remote participants, refreshed on connect/disconnect and track subscription events.
Retourne: ValueListenable<List<RemoteParticipant>> — Current remote participants.
eventsgetterasyncStream<RoomEvent> get eventsRaw broadcast stream of LiveKit RoomEvents for apps that want the full firehose (track subscribed, data, metadata, speakers).
Retourne: Stream<RoomEvent> — Broadcast LiveKit event stream.
localParticipantgetterLocalParticipant? get localParticipantThe local LiveKit participant, or null before connection.
Retourne: LocalParticipant? — Local participant or null.
setMicrophonemethodasyncFuture<void> setMicrophone(bool enabled)Toggles the local mic. Refused by the SFU unless the server has granted publish.
Paramètres
enabledboolrequisWhether to enable the microphone.
Retourne: Future<void> — Completes when applied.
sendDatamethodasyncFuture<void> sendData(List<int> data, {bool reliable = true, List<String>? to})Sends a data message (chat/gifts/custom signalling) to the room, optionally to specific identities.
Paramètres
dataList<int>requisRaw bytes payload.
reliableboolPar défaut = trueUse reliable delivery.
toList<String>?Target participant identities; null broadcasts to all.
Retourne: Future<void> — Completes when sent.
leavemethodasyncFuture<void> leave()Leaves the media room but keeps the session object reusable.
Retourne: Future<void> — Completes after disconnect.
disposemethodasyncFuture<void> dispose()Tears down the session and its LiveKit room.
Retourne: Future<void> — Completes after teardown.
Call session (1:1)
1:1 audio/video call lifecycle. Caller starts, callee accepts; room capped at 2 by the engine.
UTDCallSessionclassclass UTDCallSession extends UTDSession1:1 call session (audio_call or video_call). Both parties publish. Caller start()s; callee accept()s an incoming call by id.
statuspropertyfinal ValueNotifier<UTDCallStatus?> statusListenable for the current call lifecycle status, null until a call starts.
Retourne: ValueNotifier<UTDCallStatus?> — Current call status.
callIdpropertyfinal ValueNotifier<String?> callIdListenable for the active engine call id, null until a call starts/accepts.
Retourne: ValueNotifier<String?> — Active call id.
startmethodasyncFuture<UTDCall> start({required String calleeIdentity, String? calleeName, Map<String, dynamic>? metadata})Caller side: creates the call, rings the callee, joins the call room and publishes. Returns the engine call.
Paramètres
calleeIdentityStringrequisIdentity of the user being called.
calleeNameString?Optional callee display name.
metadataMap<String, dynamic>?Optional call metadata.
Retourne: Future<UTDCall> — The created engine call.
acceptmethodasyncFuture<UTDCall> accept(String id)Callee side: accepts an incoming call by id and joins the call room, publishing mic (and camera for video).
Paramètres
idStringrequisThe incoming call id (arrives out-of-band).
Retourne: Future<UTDCall> — The accepted engine call.
ringingmethodasyncFuture<void> ringing(String id)Marks the call as ringing on the engine and updates status.
Paramètres
idStringrequisThe call id.
Retourne: Future<void> — Completes when status updated.
rejectmethodasyncFuture<void> reject(String id)Rejects an incoming call and updates status.
Paramètres
idStringrequisThe call id to reject.
Retourne: Future<void> — Completes when rejected.
busymethodasyncFuture<void> busy(String id)Marks the callee as busy for the given call and updates status.
Paramètres
idStringrequisThe call id.
Retourne: Future<void> — Completes when marked busy.
endmethodasyncFuture<void> end()Hangs up the active call, ending it server-side and disconnecting the media room. Tolerant of terminal/duplicate ends.
Retourne: Future<void> — Completes after teardown.
Live stream session & stage control
One host + unlimited viewers; host promotes any number of viewers to co-publish (no engine cap).
UTDLiveStreamSessionclassclass UTDLiveStreamSession extends UTDSessionLive stream session: one host publishes A/V, unlimited viewers subscribe, host can promote any number of co-publishers. Stage promotion is server-authoritative.
roomNamepropertyfinal String roomNameThe (un-prefixed) live stream room name.
Retourne: String — Room name.
isHostpropertyfinal bool isHostWhether this participant joined as the host.
Retourne: bool — Host flag.
stagepropertyfinal ValueNotifier<List<UTDStageMember>> stageListenable list of current stage members (host + co-publishers), refreshed via refreshStage() or stage events.
Retourne: ValueNotifier<List<UTDStageMember>> — Current stage members.
goLivemethodasyncFuture<void> goLive({bool withVideo = true})Host: begins broadcasting by publishing mic and (optionally) camera. Viewers never call this.
Paramètres
withVideoboolPar défaut = truePublish camera in addition to mic.
Retourne: Future<void> — Completes when broadcasting.
refreshStagemethodasyncFuture<void> refreshStage()Fetches the current stage roster from the engine and updates the stage listenable.
Retourne: Future<void> — Completes when stage refreshed.
inviteToStagemethodasyncFuture<void> inviteToStage(String identity)Host/admin: promotes a viewer to co-publish (audio+video). Unbounded.
Paramètres
identityStringrequisViewer identity to promote.
Retourne: Future<void> — Completes after promotion and stage refresh.
removeFromStagemethodasyncFuture<void> removeFromStage(String identity)Host/admin: removes a co-publisher back to viewer.
Paramètres
identityStringrequisCo-publisher identity to demote.
Retourne: Future<void> — Completes after demotion and stage refresh.
requestStagemethodasyncFuture<void> requestStage()Viewer: raises a hand to be promoted, notifying the host.
Retourne: Future<void> — Completes when request sent.
leaveStagemethodasyncFuture<void> leaveStage()Co-publisher: steps down to a viewer, stopping camera and mic.
Retourne: Future<void> — Completes after stepping down.
Audio room session & seat control
Seat-based speakers + listeners; taking a seat is what grants publish. Audio-only by design.
UTDAudioRoomSessionclassclass UTDAudioRoomSession extends UTDSessionAudio room session: speakers occupy seats and publish mic, listeners subscribe. Publishing is seat occupancy, not a token role. Audio-only.
roomNamepropertyfinal String roomNameThe (un-prefixed) audio room name.
Retourne: String — Room name.
isHostpropertyfinal bool isHostWhether this participant joined as the host.
Retourne: bool — Host flag.
setupSeatsmethodasyncFuture<Map<String, dynamic>> setupSeats({int? seatCount, String? seatMode, String? modeId})Host: configures the seat grid (count + free/request mode). Usually passed at join instead. Returns the raw engine response.
Paramètres
seatCountint?Number of seats.
seatModeString?Seat mode (e.g. free/request).
modeIdString?Optional mode identifier.
Retourne: Future<Map<String, dynamic>> — Raw engine seat-setup response.
getSeatsmethodasyncFuture<Map<String, dynamic>> getSeats()Fetches the current seat grid state from the engine.
Retourne: Future<Map<String, dynamic>> — Raw engine seats state.
takeSeatmethodasyncFuture<void> takeSeat(int index)Takes a seat to become a speaker; the engine grants publish, then the mic is enabled locally.
Paramètres
indexintrequisSeat index to occupy.
Retourne: Future<void> — Completes when seated and mic on.
leaveSeatmethodasyncFuture<void> leaveSeat()Leaves the seat (back to listener); the engine revokes publish and the mic is disabled.
Retourne: Future<void> — Completes when off-seat.
requestToSpeakmethodasyncFuture<void> requestToSpeak()Listener: requests to speak (in request mode); a host approves.
Retourne: Future<void> — Completes when request sent.
approveSpeakermethodasyncFuture<void> approveSpeaker(int requestId)Host: approves a pending speak request by id.
Paramètres
requestIdintrequisPending request id to approve.
Retourne: Future<void> — Completes when approved.
inviteToSpeakmethodasyncFuture<void> inviteToSpeak(String identity, {int? seatIndex})Host: invites a listener to speak, optionally targeting a specific seat.
Paramètres
identityStringrequisListener identity to invite.
seatIndexint?Optional target seat index.
Retourne: Future<void> — Completes when invite sent.
Room wrapper & media
Headless wrapper around a LiveKit Room exposing connection state, media toggles, and the event stream.
UTDRoomconstructorUTDRoom({bool adaptiveStream = true, bool dynacast = true})Headless wrapper around a LiveKit Room. Owns the connection lifecycle and surfaces state as ValueListenables + a broadcast event stream. Never elevates its own permissions.
Paramètres
adaptiveStreamboolPar défaut = trueEnable LiveKit adaptive stream.
dynacastboolPar défaut = trueEnable LiveKit dynacast.
roompropertyfinal Room roomThe underlying LiveKit Room instance.
Retourne: Room — LiveKit room.
eventsgetterasyncStream<RoomEvent> get eventsRaw LiveKit room events for apps that want the full firehose (track subscribed, data, etc.).
Retourne: Stream<RoomEvent> — Broadcast LiveKit event stream.
connectmethodasyncFuture<void> connect(String url, String token, {bool autoSubscribe = true})Connects to the LiveKit room with the given url and token, wires event listeners, and refreshes permissions/remotes.
Paramètres
urlStringrequisLiveKit ws URL.
tokenStringrequisLiveKit JWT.
autoSubscribeboolPar défaut = trueAuto-subscribe to remote tracks.
Retourne: Future<void> — Completes when connected.
setMicrophonemethodasyncFuture<void> setMicrophone(bool enabled)Enables/disables the local microphone. Refused by the SFU unless the server granted publish.
Paramètres
enabledboolrequisWhether to enable the mic.
Retourne: Future<void> — Completes when applied.
setCameramethodasyncFuture<void> setCamera(bool enabled)Enables/disables the local camera. Refused by the SFU unless the server granted publish.
Paramètres
enabledboolrequisWhether to enable the camera.
Retourne: Future<void> — Completes when applied.
sendDatamethodasyncFuture<void> sendData(List<int> data, {bool reliable = true, List<String>? destinationIdentities})Sends a data message to the room (chat/gifts/custom signalling), optionally to specific identities.
Paramètres
dataList<int>requisRaw bytes payload.
reliableboolPar défaut = trueUse reliable delivery.
destinationIdentitiesList<String>?Target identities; null broadcasts.
Retourne: Future<void> — Completes when sent.
Models & enums
Data types and enums returned by the engine: stream types, call lifecycle, token responses, errors.
UTDStreamTypeenumenum UTDStreamType { audioCall, videoCall, liveStream, audioRoom }The four canonical product types (audioCall, videoCall, liveStream, audioRoom). Each carries a wire value and isCall/hasVideo helpers.
UTDCallStatusenumenum UTDCallStatus { initiating, ringing, active, ended, rejected, busy, missed }Lifecycle of a 1:1 call mirroring engine call states: initiating, ringing, active, ended, rejected, busy, missed.
UTDTokenResponseclassclass UTDTokenResponse { final String token; final String url; final String roomName; final String userToken; final String? wsUrl; }Response of a room token mint (POST /api/v1/token), carrying the LiveKit token+url, room name, engine bearer user_token, and optional wsUrl.
UTDCallclassclass UTDCall { final String callId; final UTDCallStatus status; final UTDStreamType type; ... }A 1:1 call as returned by the engine, carrying callId, status, type, and optional roomName/token/url/callee+caller identities.
UTDStageMemberclassclass UTDStageMember { final String identity; final String? name; final String? role; }A member of a live_stream stage (the unbounded publisher set), with identity and optional name/role.
UTDStreamExceptionclassclass UTDStreamException implements Exception { final String message; final int? statusCode; final String? code; }All SDK errors surface as this, carrying the engine HTTP statusCode + message + optional code; exposes isBanned (403) and isRateLimited (429).
Prêt à construire avec UTD ?
Créez votre compte, alimentez votre portefeuille principal et activez les services dont vous avez besoin.