Azure/azure-sdk-for-java com.azure+azure-ai-voicelive_1.1.0-beta.1

on GitHub

1.1.0-beta.1 (2026-06-02)

Features Added

• WebRTC SDP negotiation for browser/native WebRTC clients:
  • New ClientEventRtcCallSdpCreate (rtc.call.sdp.create) to send an SDP offer with an optional initial VoiceLiveSessionOptions
  • New ServerEventRtcCallSdpCreated (rtc.call.sdp.created) carrying the SDP answer and rtcCallId
  • New ServerEventRtcCallError (rtc.call.error) with structured RtcCallErrorDetails (type/code/message) and originating operation
• Streaming input text into a conversation item:
  • New ClientEventInputTextDelta (input_text.delta) and ClientEventInputTextDone (input_text.done), each with optional contentIndex
• Output audio buffer playback lifecycle:
  • New ServerEventOutputAudioBufferStarted (output_audio_buffer.started) and ServerEventOutputAudioBufferStopped (output_audio_buffer.stopped), each exposing responseId
• Smart audio-based end-of-turn detection: new SmartEndOfTurnDetection (smart_end_of_turn_detection) EouDetection variant with thresholdLevel and timeoutMs; new EouDetectionModel.SMART_END_OF_TURN_DETECTION value
• Echo cancellation reference source: AudioEchoCancellation is now @Fluent and exposes referenceSource (new EchoCancellationReferenceSource enum with SERVER / CLIENT) and channels for stereo input where channel 1 is the client-supplied echo reference
• Azure realtime native voices: new AzureRealtimeNativeVoice (type azure-realtime-native) and AzureRealtimeNativeVoiceName expandable enum (AARTI, ANDREW, AVA, DENISE, DIYA, ELSA, FLORIAN, FRANCISCA, MEERA, XIAOXIAO, YUNXI, XIMENA) for use with the azure-realtime model
• Parallel tool calls: VoiceLiveSessionOptions.setParallelToolCalls(Boolean) / isParallelToolCalls() (and matching getter on VoiceLiveSessionResponse) to control whether the model may invoke tools in parallel
• Hosted agent invocation passthrough:
  • ResponseCreateParams.setInvokeInput(Map<String, BinaryData>) to attach input data for a hosted agent invocation on a single response request (preview)
  • New ServerEventResponseInvocationDelta (response.invocation.delta) passes through non-speech SSE events from the hosted agent

Other Changes

• Regenerated against 2026-06-01-preview VoiceLive TypeSpec. The default service API version used by VoiceLiveClientBuilder is unchanged (V2026_04_10, GA).