2.9.2

📦 Stable Release

Changes since 2.9.1

🚀 Features and enhancements

• Relax resource requirements for audio analysis providers (by @marcelveldt in #4249)
• Relax 'Maximum' buffer tier to 7GB RAM (by @marcelveldt in #4268)

🐛 Bugfixes

• Alexa: Fix restore saved session after aiohttp update (by @Joshi425 in #4181)
• Add correct playlog information when retrieving audiobooks in audiobooks controller (by @fmunkes in #4199)
• Fix Cast Group mDNS for Nest Mini stereo pairs (cast_port/leader rename) (by @goodlucknow in #4224)
• Skip multichannel files in AcoustID scan instead of crashing (by @OzGav in #4230)
• Fix progress report when transitioning from idle or paused (by @fmunkes in #4236)
• Fix image download from CDNs that reject our User-Agent (by @OzGav in #4243)
• Backfill missing album on provider album tracks (by @OzGav in #4244)
• Fix standard crossfade falling back to a hard cut on some tracks (by @marcelveldt in #4253)
• Bound audio-analysis CPU usage and silence NNPACK spam on ARM (by @marcelveldt in #4257)
• Record explicit album/artist/track plays as user-initiated (by @chrisuthe in #4260)
• Release drained audio buffers in the inactivity monitor (by @marcelveldt in #4294)

🧰 Maintenance and dependency bumps

5 changes

• Vectorize weighted distance in the sonic similarity provider (by @marcelveldt in #4203)
• Bump aiohttp from 3.14.0 to 3.14.1 (by @dependabot[bot] in #4241)
• Type ytmusic search filter as a Literal (unblock ytmusicapi 1.12.1 mypy) (by @OzGav in #4245)
• Accept Python 3.14 syntax in backports without reformatting stable (by @marcelveldt in #4256)
• Make SQLite page-cache and mmap RAM-aware (by @marcelveldt in #4293)

🙇 Thanks to our contributors

Special thanks to the following contributors who helped with this release:

@Joshi425, @OzGav, @chrisuthe, @fmunkes, @goodlucknow, @marcelveldt