fix: #199#238
Conversation
|
I'm on 15.1 beta does that work for you? |
|
@Lunchb0ne yes, although if you come for the bounty just testing the pr is $50 (since i fixed the issue myself) |
|
Not sure how should I claim this. Do I create a PR without changes or what? |
|
/tip $50 @kunokareal |
|
🎉🎈 @kunokareal has been awarded $50! 🎈🎊 |
|
ETA: will merge tomorrow |
This is my run on |
|
@Lunchb0ne thanks does this vision error happen on main branch? usually vision can show a few errors on some windows but should stop quickly or be infrequent also do you have idea what is this untitled window? cursor tab maybe? regarding the audio error i will not merge then - will do more testing with @m13v next week |
|
Here's from the latest on main. My guess for the Unknown window is Arc Browser, that is the only other window I have opened. |
|
Oh How do you build? Wonder what's special in your environment that cause this |
|
I don't think I'm doing anything differently. Using the latest stable rust toolchain on macos 15.1 Do you reckon I can get any other logs to help? |
|
@m13v can you test this? if it work for you lets resolve conflicts & merge also now anyone can record audio output using virtual device instead of display by creating aggregate audio device
typically need thing like https://github.com/ExistentialAudio/BlackHole though |
scoped to the LLM-reference block only. cross-checked against the codebase, the brain, and the in-flight pricing launch (website #238 / app #3846): - pricing: drop removed lifetime/$400/$600/$39; use launch tiers Standard $25/mo, Pro $50/seat/mo, Enterprise $150/seat/mo. lifetime grandfathered, no new lifetime sales. core engine + CLI stay free/OSS. - ai search: SQLite FTS5 keyword search, not "semantic embeddings" (embeddings exist only for speaker-id; content embeddings are a future TODO) - built-in pipes: replace nonexistent obsidian-sync/reminders/idea-tracker with the actually-shipped pipes (meeting-summary, day-recap, etc.) - audio: Whisper Large-V3-Turbo locally / Deepgram cloud - mcp: pin @latest to match the install snippet - faq + comparison table: one-time/lifetime -> subscription reality Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
scoped to the LLM-reference block only. cross-checked against the codebase, the brain, and the in-flight pricing launch (website #238 / app #3846): - pricing: drop removed lifetime/$400/$600/$39; use launch tiers Standard $25/mo, Pro $50/seat/mo, Enterprise $150/seat/mo. lifetime grandfathered, no new lifetime sales. core engine + CLI stay free/OSS. - ai search: SQLite FTS5 keyword search, not "semantic embeddings" (embeddings exist only for speaker-id; content embeddings are a future TODO) - built-in pipes: replace nonexistent obsidian-sync/reminders/idea-tracker with the actually-shipped pipes (meeting-summary, day-recap, etc.) - audio: Whisper Large-V3-Turbo locally / Deepgram cloud - mcp: pin @latest to match the install snippet - faq + comparison table: one-time/lifetime -> subscription reality Co-authored-by: Louis Beaumont <louis@screenpi.pe> Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>



looking for someone on macos beta (>=15.0) to test this PR (i'm on 14.6)
how to test:
this should use your default audio output
play some youtube video or something like audiobook idk someone speaking
after 1 min you should see some transcriptions in the terminal
if this is the case can merge