v0.2.0

What's Changed

• Smaller and move transpose to the Encoder Model by @BrandonWeng in #41
• Update README.md by @BrandonWeng in #42
• Cleanup ASR pipeline code and remove redundant paths by @Alex-Wengg in #40
• Add Voice dataset (subset) benchmark for vad by @BrandonWeng in #43
• Make result structs sendable for ASR by @BrandonWeng in #46
• Make AsrModels and DiarizerModels properly Sendable by @BrandonWeng in #48
• Metal optimization for ASR by @BrandonWeng in #47
• Fixes for ASR TDT and provide example for streaming ASR by @BrandonWeng in #50
• Fix ComputeUnits override for ASR by @BrandonWeng in #54
• Fix Float16 compatibility for macOS deployment targets by @Beingpax in #55
• Clean up VAD documentation and add missing tranpose by @BrandonWeng in #56
• Swift Format by @BrandonWeng in #57
• fix: vocabulary loading ignores custom model directory by @Beingpax in #58
• Add Spokenly app to the showcase by @vadim-ahmerov in #60
• Int8 quantization + memory optimization for faster diarization by @Alex-Wengg in #59
• Update README.md by @BrandonWeng in #61
• Update claude-dispatch.yml by @BrandonWeng in #62

New Contributors

• @Beingpax made their first contribution in #55
• @vadim-ahmerov made their first contribution in #60

Full Changelog: v0.1.0...v0.2.0