chore: 0.4.9 - #128
Merged
Merged
Conversation
Whisper lists the languages it actually speaks, the live meeting can correct the language it is hearing, and the model can be chosen before the language.
The shell draws `ListeningIn` on every screen while a recording runs. The record screen drew its own copy as well, so that screen carried the sentence twice, each with a "Đổi" button — and a click landed on whichever the browser found first, which is also what stopped the full-flow suite.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Version bump only — the same nine files as #126.
In this release (#127)
Whisper lists the languages it speaks. A manifest writes
langs: ["*"]for the multilingual set, and the catalogue screen skipped any model containing a star — so the one model covering 99 languages was the only card showing none, next to Gipformer showingvi. Search had the same hole. The daemon now expands the star fromlanguages::MULTILINGUAL. The chip leads with the reader's own language:vi · en · zh · de +96.The live meeting can correct the language it is hearing.
ListeningInswaps the decoder mid-session viaretune. It worked, and was only ever rendered on the old/recordscreen — unreachable since recording started opening the note instead.Choose the model, then any language it speaks. Picking a language and letting the ranking choose the model is still the default; "by model" pins the model and offers all 99 of Whisper's languages, naming the recommended model when you overrule it.
Verification
cargo fmt --check,clippy -D warnings,cargo test --workspace, plus the same three inapps/desktop/src-tauri.tsc,lint,prettier --check, 441 unit tests. e2esettings/review/models/density/language/meeting/languagesagainst a rebuilt binary, and the new form driven against a real daemon — 101 language options for Whisper,models.live=whisper-tiny, models.language=vipersisted, override notice shown.Still open from the wider request
Per-language model routing at runtime, two models at once for bilingual, multiple translation targets, and mid-session model switching are not in this release.
runner.rsbuilds one decoder per session, warm-keyed by(model, language); bilingual needs N decoders and a per-utterance router at roughly double the CPU, andtranslate_tois sent only at session start.🤖 Generated with Claude Code