v0.3.0-alpha.4

课溯 · VeriLecture v0.3.0-alpha.4

把课堂录音变成可核对的复习重点。
Trace every key point back to the lecture.

This is the second public V3 milestone. It publishes the desktop shell for three platforms while keeping native local-ASR support and the CUDA Runtime status explicit.

这次发布能做什么

• 导入课堂录音，并保留原始音频和原始转写。
• 使用课程词库进行校准；修订和编辑生成带来源的新版本。
• 在文本整理服务获得授权后生成复习重点，并保留相关转写片段与时间点。
• 从重点或转写位置回到原音，或导出 Markdown、JSON 和纯文本结果。
• 在首次运行和设置中查看硬件、本地转写和数据授权边界。

What this release makes possible

• Import classroom audio while keeping the original recording and raw transcript available.
• Calibrate terminology with a course lexicon; corrections and edits create new versions with provenance.
• Generate review points through a consented text service and retain the related transcript segments and timestamps.
• Return from a point or transcript line to the source position, then export Markdown, JSON, or plain text.
• See hardware, local transcription, and consent boundaries during first run and in Settings.

Package and runtime status

Platform	Package	Desktop package	Native local ASR
Windows x64	NSIS installer	Published	Fun-ASR route and a representative CUDA path have corresponding validation
Linux x64	AppImage	Published	Native sidecar pending publication and validation
macOS	DMG	Published	Native sidecar pending publication and validation

The existence of a desktop package does not prove that local ASR is ready on that platform.

Documentation note

The public README and website include a generated result-screen concept mockup marked 概念示意 · 非实机截图. It explains the intended relationship between points, transcript segments, timestamps, and playback; it is not evidence of a verified Windows UI or runtime path. A real result capture remains pending the acceptance path in the screenshot brief.

Important limitations

• This is an Alpha release. Back up important recordings before installing.
• Linux and macOS packages provide the desktop application, but native local-ASR sidecars still require separate publication and validation.
• The Qwen3-ASR tiers require a compatible NVIDIA CUDA path and a separately hosted CUDA Runtime. The approximately 4.6 GB Runtime is excluded from Git and this Release; the installer keeps the download gated until a stable HTTPS artifact is published and accepted.
• Model weights are downloaded on first use and require a network connection, free disk space, and an integrity check.
• Long audio is processed as a bounded local job; durable in-job resume is not exposed in the current interface.
• Text-model features are optional, require a configured provider and explicit consent, and have narrower provider and long-document coverage while the Alpha is being expanded.
• The Windows installer is unsigned unless signing is supplied by the release environment.

Read the full list in Known Limitations.

Checksums

Each installer has a matching SHA256 checksum file. The checksum entry uses the same filename as the published asset. Verify the matching file before installing:

• Windows SHA256
• Linux SHA256
• macOS SHA256

First run

1. Read the privacy boundary and choose the consent scope.
2. Let VeriLecture scan hardware, CUDA, storage, and the model directory.
3. Choose a local transcription route that the device can run.
4. Test a short recording before importing a full lecture.

Choose Fun-ASR-Nano-2512 when no usable NVIDIA GPU is available. For Qwen, wait for a Release that explicitly says the CUDA Runtime is publicly hosted and accepted.

License and attribution

VeriLecture is released under the MIT License. Third-party licenses and upstream attribution are listed in NOTICE and THIRD_PARTY_NOTICES.md.

v0.3.0-alpha.3

VeriLecture v0.3.0-alpha.3

Windows x64, Linux x64, and macOS desktop packages

VeriLecture turns classroom audio into local, reviewable learning material. It
keeps the source recording and raw transcript available while making
corrections, vocabulary, and review points easier to revisit.

Highlights

• Windows x64 NSIS installer.
• Linux x64 AppImage.
• macOS DMG.
• Local-first audio import with support for WAV, MP3, and M4A.
• Immutable raw transcripts with versioned calibration and edits.
• Simplified Chinese and English interface support.
• Verified platform-specific checksums for every published installer.
• Native square application icons for the Linux AppImage and other bundles.

Important limitations

• This is an Alpha release. Back up important recordings before installing.
• Linux and macOS packages provide the desktop application, but their native
  local ASR runtimes still require separate publication and validation.
• The Qwen tiers require a separately hosted CUDA Runtime. Until that runtime
  is publicly available, choose Fun-ASR-Nano-2512 for the CPU path.
• Model weights are downloaded on first use rather than bundled in the
  installer.
• The Windows installer is unsigned unless a signing configuration is supplied
  by the release environment.

Checksums

Each platform installer is accompanied by a SHA256 checksum file in this
Release. Verify the matching checksum before installing a downloaded artifact.

License

VeriLecture is released under the MIT License. Third-party
licenses and upstream attribution are listed in
NOTICE and
THIRD_PARTY_NOTICES.md.

v0.3.0-alpha.1

VeriLecture / 课溯 v0.3.0-alpha.1

Pre-release · V3 public development milestone

VeriLecture V3 turns the local lecture workflow into a visible evidence chain: keep the original audio, preserve the raw transcript, create traceable corrections, and review derived learning material without losing the source.

Highlights

• Reframed the application as a Windows-first, local-first VeriLecture V3 product.
• Added first-run hardware scanning and model routing for three supported ASR tiers:
  • Fun-ASR-Nano-2512 for CPU fallback.
  • Qwen3-ASR-0.6B for a lighter CUDA path.
  • Qwen3-ASR-1.7B for the higher-quality CUDA path.
• Kept everyday users away from model/provider internals; technical routing remains in diagnostics.
• Preserved raw audio and transcripts while making corrections and calibration versioned.
• Added onboarding guidance, bilingual Simplified Chinese/English copy, privacy boundaries, and course lexicon entry points.
• Added Windows CI for frontend and Rust checks plus a tag-driven NSIS release workflow.
• Added a pinned, SHA-256-verified FFmpeg build-time fetch step so the large third-party binary is not committed to Git.

Verification evidence

• Fun-ASR-Nano CPU smoke path: completed in the local V3 acceptance workspace.
• Qwen3-ASR-1.7B: representative RTX 3060 CUDA transcription smoke test passed; see docs/evidence/gpu-smoke-report.json.
• Qwen3-ASR-0.6B: not independently retested in this milestone, following the project decision that a successful 1.7B GPU path is sufficient for the current validation scope.
• CUDA runtime chain: local Range/resume, SHA-256, ZIP64, and runtime probe checks passed against the local archive; this is not a public download acceptance test.

Important limitation: CUDA Runtime is not published yet

The Qwen installers depend on a shared CUDA runtime archive of approximately 4.6 GiB. That archive is intentionally not attached to this GitHub release, is not committed to the repository, and is not referenced by a public published registry entry yet.

Therefore:

• Qwen installation may remain unavailable on a clean installation until the runtime is hosted at a stable HTTPS address.
• This release must not be treated as proof that a new user can download and install Qwen end-to-end.
• CPU users should select Fun-ASR-Nano-2512.
• CUDA Runtime hosting, registry publication, and a clean Windows acceptance run are the next release-blocking tasks.

Other known limitations

• Text summary generation depends on a user-configured text provider and explicit consent; cloud text processing was not the focus of this release validation.
• This is an alpha build and may contain UI or layout issues on unusual Windows scaling settings.
• The Windows installer is currently unsigned unless a signing configuration is supplied by the release environment.
• Broader GPU/driver and clean-machine compatibility coverage is still required.

Upgrade and data safety

Back up important recordings before installing a pre-release. VeriLecture is designed to keep original audio and raw transcripts intact, but alpha software should not be the only copy of valuable course material.

Checksums

The CI release job publishes SHA256SUMS-v0.3.0-alpha.1.txt beside the Windows installer. Verify the checksum before installing a downloaded artifact.

The locally verified Windows NSIS artifact used for this release preparation is
课溯 · VeriLecture_0.3.0-alpha.1_x64-setup.exe (110,421,695 bytes), with SHA-256:

B59118576BCF87855088305AF32B8A7D6CC8477BDAF25191F8192A40D969F1CF