Repository navigation
feat(mobile): on-device dictation models, voice filter and correction learning on iOS - #59
Merged
Merged
Conversation
… learning on iOS Dictation on iOS and iPadOS could only use Apple's recognizer: no choice of model, no way to keep other people's speech out, and the same mishearings had to be fixed by hand every time. Settings > Microphone now offers 3 downloadable Parakeet models (Fast, Balanced, Most accurate), each gated by the device's memory and free space. "Only my voice" records a voice profile and removes other speakers before transcribing. Replacement rules can be written by hand or learned from the edits made to a dictation before it is sent. All of it runs on the device; the network is used only to download a model. The feature lives in a fork-only Expo module (modules/pseudo-dictation), a config plugin that adds the FluidAudio pod, and features/dictation, so upstream files change in 6 small places. Android and builds without the module keep the existing behavior. A cancelled download removes its files. A failed one keeps them so a retry resumes. Model: Claude Fable 5. Harness: Claude Code in PseudoCode. Co-Authored-By: Claude Code <noreply@anthropic.com>
Thread transfer impact
This comment will update automatically after the next completed run. |
Owner
Author
|
Merging with "Mobile Native Static Analysis" cancelled. GitHub did not assign a macOS runner in 2 attempts of 15 minutes each (annotation: "The job was not acquired by Runner of type hosted even after multiple attempts"), so the job never ran, and "Check" fails only because of it. The same lint ( |
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
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.
Problem
Dictation on iOS and iPadOS could only use Apple's recognizer. There was no choice of model, no way to keep other people's speech out, and the same mishearings had to be fixed by hand every time.
Fix
Fork-only feature. It is not meant for upstream.
apps/mobile/modules/pseudo-dictation), a config plugin that adds the FluidAudio pod, andapps/mobile/src/features/dictation. Upstream files change in 6 small places. Android and builds without the module keep the existing behavior.Verification
vp run check:changedexited 0.Not verified here: Apple's recognizer path (not available on the iOS 27 simulator). Screenshots from the simulator pass are kept locally and are not attached.
Known limits
Model: Claude Fable 5. Harness: Claude Code in PseudoCode.
🤖 Generated with Claude Code