v1.8.1

Download Options

Package	Description
birda-linux-x64-v1.8.1.tar.gz	Linux CPU-only
birda-windows-x64-v1.8.1.zip	Windows CPU-only
birda-macos-arm64-v1.8.1-signed.tar.gz	macOS Apple Silicon (Signed & Notarized)
birda-linux-x64-cuda-v1.8.1.tar.gz	Linux with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.8.1.zip	Windows with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.8.1-setup.exe	Windows GPU installer (adds to PATH)
birda-linux-x64-embed-v1.8.1.tar.gz	Linux embeddable (CLI + ONNX Runtime CPU)
birda-windows-x64-embed-v1.8.1.zip	Windows embeddable (CLI + ONNX Runtime CPU)
birda-macos-arm64-embed-v1.8.1.tar.gz	macOS embeddable (Signed, CLI + ONNX Runtime CPU)
birda-linux-x64-bin-v1.8.1.tar.gz	Linux binary-only (for updates)
birda-linux-x64-cuda-bin-v1.8.1.tar.gz	Linux CUDA binary-only (for updates)
birda-windows-x64-bin-v1.8.1.zip	Windows binary-only (for updates)
birda-windows-x64-cuda-bin-v1.8.1.zip	Windows CUDA binary-only (for updates)
birda-macos-arm64-bin-v1.8.1.tar.gz	macOS binary-only (for updates)
birda-cuda-libs-linux-x64-v1.8.1.tar.gz	Linux CUDA libraries only
birda-cuda-libs-windows-x64-v1.8.1.zip	Windows CUDA libraries only
manifest.json	Version manifest for GUI integration

macOS Installation

Extract the tar.gz and copy birda to your PATH. The binary is signed and notarized by Apple with hardened runtime enabled - no Gatekeeper warnings.

GPU Builds

GPU builds include bundled ONNX Runtime, CUDA, and cuDNN libraries - no separate installation required.

Windows Installer: Use the -setup.exe for automatic installation with optional PATH configuration.

Portable: Extract ZIP/tar.gz and run birda directly.

Requirements: NVIDIA GPU with CUDA compute capability 5.0+

TensorRT Support (Optional)

For maximum GPU performance (~2x speedup over CUDA), install TensorRT separately:

1. Download TensorRT for CUDA 12.x from NVIDIA TensorRT
2. Copy the TensorRT DLLs/libs to the birda installation directory
3. Run with --tensorrt flag

See THIRD_PARTY_LICENSES.txt for bundled library licenses.

What's Changed

• chore: update third-party licenses by @github-actions[bot] in #205

Full Changelog: v1...v1.8.1

v1.8.0

Download Options

Package	Description
birda-linux-x64-v1.8.0.tar.gz	Linux CPU-only
birda-windows-x64-v1.8.0.zip	Windows CPU-only
birda-macos-arm64-v1.8.0-signed.tar.gz	macOS Apple Silicon (Signed & Notarized)
birda-linux-x64-cuda-v1.8.0.tar.gz	Linux with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.8.0.zip	Windows with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.8.0-setup.exe	Windows GPU installer (adds to PATH)
birda-linux-x64-embed-v1.8.0.tar.gz	Linux embeddable (CLI + ONNX Runtime CPU)
birda-windows-x64-embed-v1.8.0.zip	Windows embeddable (CLI + ONNX Runtime CPU)
birda-macos-arm64-embed-v1.8.0.tar.gz	macOS embeddable (Signed, CLI + ONNX Runtime CPU)
birda-cuda-libs-linux-x64-v1.8.0.tar.gz	Linux CUDA libraries only
birda-cuda-libs-windows-x64-v1.8.0.zip	Windows CUDA libraries only
manifest.json	Version manifest for GUI integration

macOS Installation

Extract the tar.gz and copy birda to your PATH. The binary is signed and notarized by Apple with hardened runtime enabled - no Gatekeeper warnings.

GPU Builds

GPU builds include bundled ONNX Runtime, CUDA, and cuDNN libraries - no separate installation required.

Windows Installer: Use the -setup.exe for automatic installation with optional PATH configuration.

Portable: Extract ZIP/tar.gz and run birda directly.

Requirements: NVIDIA GPU with CUDA compute capability 5.0+

TensorRT Support (Optional)

For maximum GPU performance (~2x speedup over CUDA), install TensorRT separately:

1. Download TensorRT for CUDA 12.x from NVIDIA TensorRT
2. Copy the TensorRT DLLs/libs to the birda installation directory
3. Run with --tensorrt flag

See THIRD_PARTY_LICENSES.txt for bundled library licenses.

What's Changed

• deps: bump quinn-proto from 0.11.13 to 0.11.14 by @dependabot[bot] in #169
• deps: bump lz4_flex from 0.12.0 to 0.12.1 by @dependabot[bot] in #173
• deps: bump rustls-webpki from 0.103.9 to 0.103.10 by @dependabot[bot] in #177
• deps: bump toml from 1.0.4+spec-1.1.0 to 1.1.2+spec-1.1.0 by @dependabot[bot] in #188
• deps: bump clap from 4.5.60 to 4.6.0 by @dependabot[bot] in #175
• deps: bump tracing-subscriber from 0.3.22 to 0.3.23 by @dependabot[bot] in #174
• deps: bump assert_cmd from 2.1.2 to 2.2.0 by @dependabot[bot] in #172
• deps: bump tempfile from 3.26.0 to 3.27.0 by @dependabot[bot] in #171
• deps: bump arrow from 58.0.0 to 58.1.0 by @dependabot[bot] in #179
• deps: bump parquet from 58.0.0 to 58.1.0 by @dependabot[bot] in #178
• [codex] Add AGENTS.md symlink by @maciej in #183
• [codex] Fail fast when ONNX Runtime is missing by @maciej in #185
• fix(ci): update bundled ONNX Runtime to 1.24.2 by @tphakala in #194
• deps: bump rubato 2.0, audioadapter-buffers 3.0, reqwest 0.13 by @tphakala in #195
• chore: update third-party licenses by @github-actions[bot] in #197
• feat: cross-model range filtering for perch-v2 by @tphakala in #198
• fix: negative coords parsing and cross-model range filter by @tphakala in #201

New Contributors

• @maciej made their first contribution in #183

Full Changelog: v1...v1.8.0

v1.7.0

Download Options

Package	Description
birda-linux-x64-v1.7.0.tar.gz	Linux CPU-only
birda-windows-x64-v1.7.0.zip	Windows CPU-only
birda-macos-arm64-v1.7.0-signed.tar.gz	macOS Apple Silicon (Signed & Notarized)
birda-linux-x64-cuda-v1.7.0.tar.gz	Linux with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.7.0.zip	Windows with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.7.0-setup.exe	Windows GPU installer (adds to PATH)
birda-linux-x64-embed-v1.7.0.tar.gz	Linux embeddable (CLI + ONNX Runtime CPU)
birda-windows-x64-embed-v1.7.0.zip	Windows embeddable (CLI + ONNX Runtime CPU)
birda-macos-arm64-embed-v1.7.0.tar.gz	macOS embeddable (Signed, CLI + ONNX Runtime CPU)
birda-cuda-libs-linux-x64-v1.7.0.tar.gz	Linux CUDA libraries only
birda-cuda-libs-windows-x64-v1.7.0.zip	Windows CUDA libraries only
manifest.json	Version manifest for GUI integration

macOS Installation

Extract the tar.gz and copy birda to your PATH. The binary is signed and notarized by Apple with hardened runtime enabled - no Gatekeeper warnings.

GPU Builds

GPU builds include bundled ONNX Runtime, CUDA, and cuDNN libraries - no separate installation required.

Windows Installer: Use the -setup.exe for automatic installation with optional PATH configuration.

Portable: Extract ZIP/tar.gz and run birda directly.

Requirements: NVIDIA GPU with CUDA compute capability 5.0+

TensorRT Support (Optional)

For maximum GPU performance (~2x speedup over CUDA), install TensorRT separately:

1. Download TensorRT for CUDA 12.x from NVIDIA TensorRT
2. Copy the TensorRT DLLs/libs to the birda installation directory
3. Run with --tensorrt flag

See THIRD_PARTY_LICENSES.txt for bundled library licenses.

What's Changed

• deps: bump toml from 1.0.1+spec-1.1.0 to 1.0.3+spec-1.1.0 by @dependabot[bot] in #141
• deps: bump clap from 4.5.58 to 4.5.59 by @dependabot[bot] in #140
• deps: bump ctrlc from 3.5.1 to 3.5.2 by @dependabot[bot] in #139
• ci: bump apple-actions/import-codesign-certs from 3 to 6 by @dependabot[bot] in #138
• chore: update third-party licenses by @github-actions[bot] in #136
• deps: bump clap from 4.5.59 to 4.5.60 by @dependabot[bot] in #148
• deps: bump chrono from 0.4.43 to 0.4.44 by @dependabot[bot] in #147
• deps: bump serial_test from 3.3.1 to 3.4.0 by @dependabot[bot] in #145
• deps: bump parquet from 57.3.0 to 58.0.0 by @dependabot[bot] in #146
• deps: bump arrow from 57.3.0 to 58.0.0 by @dependabot[bot] in #144
• chore: update third-party licenses by @github-actions[bot] in #154
• deps: update transitive dependencies by @tphakala in #156
• chore: update third-party licenses by @github-actions[bot] in #155
• chore: update third-party licenses by @github-actions[bot] in #158
• feat: add models remove command by @tphakala in #160
• feat: complete model management CLI commands (closes #159) by @tphakala in #161
• feat: add embeddable CLI artifacts and manifest for GUI bundling by @tphakala in #162

Full Changelog: v1...v1.7.0

v1.6.1

Download Options

Package	Description
birda-linux-x64-v1.6.1.tar.gz	Linux CPU-only
birda-windows-x64-v1.6.1.zip	Windows CPU-only
birda-macos-arm64-v1.6.1-signed.tar.gz	macOS Apple Silicon (Signed & Notarized)
birda-linux-x64-cuda-v1.6.1.tar.gz	Linux with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.6.1.zip	Windows with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.6.1-setup.exe	Windows GPU installer (adds to PATH)

macOS Installation

Extract the tar.gz and copy birda to your PATH. The binary is signed and notarized by Apple with hardened runtime enabled - no Gatekeeper warnings.

GPU Builds

GPU builds include bundled ONNX Runtime, CUDA, and cuDNN libraries - no separate installation required.

Windows Installer: Use the -setup.exe for automatic installation with optional PATH configuration.

Portable: Extract ZIP/tar.gz and run birda directly.

Requirements: NVIDIA GPU with CUDA compute capability 5.0+

TensorRT Support (Optional)

For maximum GPU performance (~2x speedup over CUDA), install TensorRT separately:

1. Download TensorRT for CUDA 12.x from NVIDIA TensorRT
2. Copy the TensorRT DLLs/libs to the birda installation directory
3. Run with --tensorrt flag

See THIRD_PARTY_LICENSES.txt for bundled library licenses.

Full Changelog: v1...v1.6.1

v1.6.0

Download Options

Package	Description
birda-linux-x64-v1.6.0.tar.gz	Linux CPU-only
birda-windows-x64-v1.6.0.zip	Windows CPU-only
birda-macos-arm64-v1.6.0.dmg	macOS Apple Silicon (Signed & Notarized DMG)
birda-macos-arm64-v1.6.0-signed.tar.gz	macOS Apple Silicon (Signed & Notarized tar.gz)
birda-linux-x64-cuda-v1.6.0.tar.gz	Linux with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.6.0.zip	Windows with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.6.0-setup.exe	Windows GPU installer (adds to PATH)

macOS Installation

DMG (Recommended): Download the .dmg file and drag to Applications folder. The app is signed and notarized by Apple - no Gatekeeper warnings.

tar.gz: For command-line users and package managers (Homebrew). Also signed and notarized.

Both formats include the same signed and notarized binary with hardened runtime enabled for maximum security.

GPU Builds

GPU builds include bundled ONNX Runtime, CUDA, and cuDNN libraries - no separate installation required.

Windows Installer: Use the -setup.exe for automatic installation with optional PATH configuration.

Portable: Extract ZIP/tar.gz and run birda directly.

Requirements: NVIDIA GPU with CUDA compute capability 5.0+

TensorRT Support (Optional)

For maximum GPU performance (~2x speedup over CUDA), install TensorRT separately:

1. Download TensorRT for CUDA 12.x from NVIDIA TensorRT
2. Copy the TensorRT DLLs/libs to the birda installation directory
3. Run with --tensorrt flag

See THIRD_PARTY_LICENSES.txt for bundled library licenses.

What's Changed

• feat: Add macOS code signing and notarization by @tphakala in #120
• fix: exclude CoreML from auto-selection on macOS by @tphakala in #121
• feat: TensorRT library detection with graceful fallback by @tphakala in #122
• fix: address code review findings from TensorRT detection feature by @tphakala in #123
• feat: add CUDA library detection with optimized shared infrastructure by @tphakala in #124
• test: add comprehensive tests for providers command by @tphakala in #129
• refactor: improve providers command tests by @tphakala in #131
• chore: update third-party licenses by @github-actions[bot] in #126
• refactor: improve Parquet implementation code quality by @tphakala in #132
• feat: add model and execution provider-specific default batch sizes by @tphakala in #134
• feat: add batch size guardrails to prevent GPU memory exhaustion by @tphakala in #135

Full Changelog: v1...v1.6.0

v1.5.0

Download Options

Package	Description
birda-linux-x64-v1.5.0.tar.gz	Linux CPU-only
birda-windows-x64-v1.5.0.zip	Windows CPU-only
birda-macos-arm64-v1.5.0.tar.gz	macOS Apple Silicon CPU-only
birda-linux-x64-cuda-v1.5.0.tar.gz	Linux with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.5.0.zip	Windows with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.5.0-setup.exe	Windows GPU installer (adds to PATH)

GPU Builds

GPU builds include bundled ONNX Runtime, CUDA, and cuDNN libraries - no separate installation required.

Windows Installer: Use the -setup.exe for automatic installation with optional PATH configuration.

Portable: Extract ZIP/tar.gz and run birda directly.

Requirements: NVIDIA GPU with CUDA compute capability 5.0+

TensorRT Support (Optional)

For maximum GPU performance (~2x speedup over CUDA), install TensorRT separately:

1. Download TensorRT for CUDA 12.x from NVIDIA TensorRT
2. Copy the TensorRT DLLs/libs to the birda installation directory
3. Run with --tensorrt flag

See THIRD_PARTY_LICENSES.txt for bundled library licenses.

What's Changed

• chore: update dependencies by @tphakala in #111
• feat: implement dual output mode for directory analysis by @tphakala in #114
• chore: update third-party licenses by @github-actions[bot] in #113
• deps: bump toml from 0.9.12+spec-1.1.0 to 1.0.0+spec-1.1.0 by @dependabot[bot] in #115
• deps: bump ctrlc from 3.5.1 to 3.5.2 by @dependabot[bot] in #116

Full Changelog: v1...v1.5.0

v1.4.2

Download Options

Package	Description
birda-linux-x64-v1.4.2.tar.gz	Linux CPU-only
birda-windows-x64-v1.4.2.zip	Windows CPU-only
birda-macos-arm64-v1.4.2.tar.gz	macOS Apple Silicon CPU-only
birda-linux-x64-cuda-v1.4.2.tar.gz	Linux with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.4.2.zip	Windows with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.4.2-setup.exe	Windows GPU installer (adds to PATH)

GPU Builds

GPU builds include bundled ONNX Runtime, CUDA, and cuDNN libraries - no separate installation required.

Windows Installer: Use the -setup.exe for automatic installation with optional PATH configuration.

Portable: Extract ZIP/tar.gz and run birda directly.

Requirements: NVIDIA GPU with CUDA compute capability 5.0+

TensorRT Support (Optional)

For maximum GPU performance (~2x speedup over CUDA), install TensorRT separately:

1. Download TensorRT for CUDA 12.x from NVIDIA TensorRT
2. Copy the TensorRT DLLs/libs to the birda installation directory
3. Run with --tensorrt flag

See THIRD_PARTY_LICENSES.txt for bundled library licenses.

Full Changelog: v1...v1.4.2

v1.4.1

Download Options

Package	Description
birda-linux-x64-v1.4.1.tar.gz	Linux CPU-only
birda-windows-x64-v1.4.1.zip	Windows CPU-only
birda-macos-arm64-v1.4.1.tar.gz	macOS Apple Silicon CPU-only
birda-linux-x64-cuda-v1.4.1.tar.gz	Linux with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.4.1.zip	Windows with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.4.1-setup.exe	Windows GPU installer (adds to PATH)

GPU Builds

GPU builds include bundled ONNX Runtime, CUDA, and cuDNN libraries - no separate installation required.

Windows Installer: Use the -setup.exe for automatic installation with optional PATH configuration.

Portable: Extract ZIP/tar.gz and run birda directly.

Requirements: NVIDIA GPU with CUDA compute capability 5.0+

TensorRT Support (Optional)

For maximum GPU performance (~2x speedup over CUDA), install TensorRT separately:

1. Download TensorRT for CUDA 12.x from NVIDIA TensorRT
2. Copy the TensorRT DLLs/libs to the birda installation directory
3. Run with --tensorrt flag

See THIRD_PARTY_LICENSES.txt for bundled library licenses.

What's Changed

• feat: add BSG Finnish Birds v4.4 model and auto-update registry by @tphakala in #109

Full Changelog: v1...v1.4.1

v1.4.0

Download Options

Package	Description
birda-linux-x64-v1.4.0.tar.gz	Linux CPU-only
birda-windows-x64-v1.4.0.zip	Windows CPU-only
birda-macos-arm64-v1.4.0.tar.gz	macOS Apple Silicon CPU-only
birda-linux-x64-cuda-v1.4.0.tar.gz	Linux with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.4.0.zip	Windows with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.4.0-setup.exe	Windows GPU installer (adds to PATH)

GPU Builds

GPU builds include bundled ONNX Runtime, CUDA, and cuDNN libraries - no separate installation required.

Windows Installer: Use the -setup.exe for automatic installation with optional PATH configuration.

Portable: Extract ZIP/tar.gz and run birda directly.

Requirements: NVIDIA GPU with CUDA compute capability 5.0+

TensorRT Support (Optional)

For maximum GPU performance (~2x speedup over CUDA), install TensorRT separately:

1. Download TensorRT for CUDA 12.x from NVIDIA TensorRT
2. Copy the TensorRT DLLs/libs to the birda installation directory
3. Run with --tensorrt flag

See THIRD_PARTY_LICENSES.txt for bundled library licenses.

Full Changelog: v1...v1.4.0

v1.3.0

Download Options

Package	Description
birda-linux-x64-v1.3.0.tar.gz	Linux CPU-only
birda-windows-x64-v1.3.0.zip	Windows CPU-only
birda-linux-x64-cuda-v1.3.0.tar.gz	Linux with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.3.0.zip	Windows with GPU support (bundled CUDA)
birda-windows-x64-cuda-v1.3.0-setup.exe	Windows GPU installer (adds to PATH)

GPU Builds

GPU builds include bundled ONNX Runtime, CUDA, and cuDNN libraries - no separate installation required.

Windows Installer: Use the -setup.exe for automatic installation with optional PATH configuration.

Portable: Extract ZIP/tar.gz and run birda directly.

Requirements: NVIDIA GPU with CUDA compute capability 5.0+

TensorRT Support (Optional)

For maximum GPU performance (~2x speedup over CUDA), install TensorRT separately:

1. Download TensorRT for CUDA 12.x from NVIDIA TensorRT
2. Copy the TensorRT DLLs/libs to the birda installation directory
3. Run with --tensorrt flag

See THIRD_PARTY_LICENSES.txt for bundled library licenses.

What's Changed

• feat(clipper): add clip subcommand for audio extraction by @tphakala in #61
• feat: add JSON output mode for programmatic integration by @tphakala in #62
• fix: pad partial batches for TensorRT batch size alignment by @tphakala in #64

Full Changelog: v1.2.5...v1.3.0