Changelog 1.1.2

✨ New Features

• Audio File Export (MP3, OGG, WAV):
  • Added asynchronous audio export to VoiceRecordingService: exportRecording(UUID/VoiceRecording, AudioExportFormat, [fileName]).
  • Converts audio into MP3, OGG Vorbis, or standard PCM WAV into plugins/DreamVoice/exports/.
  • Decodes Opus speech stream directly and accurately using sequential stream reconstruction with natural conversational pause detection, eliminating distortion and crackling.
  • Added command: /record export <id> <mp3|ogg|wav> [fileName] with tab-completion, custom destination file naming, and direct file path feedback.

🐛 Bug Fixes & Audio Improvements

• Fixed VoiceWallOcclusionEvent Async Dispatch Exception:

  • Resolved java.lang.IllegalStateException: VoiceWallOcclusionEvent may only be triggered synchronously occurring when VoiceWallServiceImpl.processEntitySoundPacket dispatched acoustic occlusion checks on Simple Voice Chat's audio processing thread (Server$ProcessThread).
  • VoiceWallOcclusionEvent now explicitly calls super(true) in its constructor to mark the Bukkit event as asynchronous, allowing Paper's PaperEventManager to dispatch it safely across worker threads without exceptions.
  • Hardened EntitySoundPacketEvent with an explicit asynchronous constructor (super(true)) to prevent similar thread check violations.

• Restored Speaker Audio Feedback for Nearby Senders:

  • Fixed an issue where players speaking into or next to a 3D speaker could not hear their own voice coming out of the speaker.
  • Removed sending-player exclusions (!sp.getUuid().equals(senderUuid) in broadcastToDedicatedChannels and if (listener.getUniqueId().equals(senderUuid)) return; in processSingleListenerStream).
  • Players standing within the speaker's acoustic range (distance) and line-of-sight now naturally hear their voice projected from the speaker's 3D locational coordinates.
  • Integrated speaker.getFilter() listener validation within processSingleListenerStream to ensure custom speaker recipient predicates are consistently respected across both direct streaming and DSP-filtered pipelines.

• Complete DSP Voice Filter Overhaul:

  • Radio (RadioVoiceFilter): Replaced 10-bit stair-step quantization (which caused harsh clicking/crackling) and 1st-order IIR with a cascaded 4th-order Butterworth bandpass (480 Hz – 2600 Hz, 24 dB/oct), mid-range presence boost (+4 dB at 1.5 kHz), warm hyperbolic tangent ($\tanh$) analog overdrive, and subtle RF carrier hiss (-42 dB).
  • Telephone (TelephoneVoiceFilter): Implemented 2nd-order Butterworth bandpass (350 Hz – 3400 Hz) with vintage carbon capsule asymmetrical saturation.
  • Alien (AlienVoiceFilter): Implemented linear fractional-delay interpolation on the sinus modulation line, eliminating sample-skipping clicks.
  • Ghost (GhostVoiceFilter): Added smooth linear delay interpolation on dual chorusing delay lines with soft-knee limiting to eliminate crackling.
  • Helium & Deep (HeliumVoiceFilter, DeepVoiceFilter): Completely redesigned pitch-shifting using 40ms grains with Hann windowing (ensuring constant unity overlap-add gain) and linear interpolation, eliminating the 50 Hz buzz and window boundary clicks.
  • Cave (CaveVoiceFilter): Rebalanced wet/dry mix (0.68/0.32) and added soft limiting to prevent reverb buffer overflow and distortion on loud speech.
  • Megaphone (MegaphoneVoiceFilter): 2nd-order horn bandpass (600 Hz – 3400 Hz) with +5 dB acoustic throat resonance at 1100 Hz, analog saturation, and normalized slapback echo.
  • Robot (RobotVoiceFilter): Normalized carrier synthesis to strictly $\le 1.0$ and tuned ring modulation dry/wet balance.
  • Disguise (DisguiseVoiceFilter): Fixed buffer index clamping bug, isolated states per player UUID to prevent cross-player modulation bleeding, and added linear interpolation.