Version 0.3.0a2

Released December 31st, 2025

Added

Additional format options for yt-dlp metadata extraction; slightly faster YouTube metadata lookup.
New metadata property from YouTubeAudioSource for UX when developing front-end reponses.

Released December 30th, 2025

Reconnection logic in the case that the voice server is deallocated.
PRIORITY audio playing status - Configurable via AudioPlayer.set_priority.
VoiceWarningEvent now dispatched for high packet loss or jitter.
New YouTubeAudioSource (see note below); introduces the yt-dlp dependency.
FFmpeg now reports errors instead of failing silently.

VoiceConnection.gateway_latency renamed to latency. This simplifies the API as additional *_latency properties will not be added.

Bug in AudioPlayer.set_volume where volume could be set to None instead of falling back to connection/client configuration.
AudioPlayer now waits for audio to be ready before setting speaking state to True and dispatching AudioBeginEvent.

YouTubeAudioSource is a quality-of-life bridge that resolves YouTube URLs into direct media URLs and behaves similarly to URLAudioSource.
This source relies on yt-dlp, which depends on YouTube's undocumented internal APIs. YouTube frequently changes these internals, which may break yt-dlp without notice.
As a result, YouTubeAudioSource is best-effort and not guaranteed to remain functional long-term.
If failures occur, updating yt-dlp is usually required. Issues related to YouTube extraction should be reported upstream to the yt-dlp project.
hikari-wave pins yt-dlp to a known working version for stability. Requests to update this version should be made via an issue.

This release moves hikari-wave from alpha to beta status.