[johnmatthewtennant]

Summary

A fresh Parakeet install fails with "local voice asset archive omitted a pinned file", so a new user can't turn on Voice with local speech recognition. The sherpa-onnx release archive names its entries ./sherpa-onnx-nemo-parakeet_tdt_ctc_110m-en-36000-int8/model.int8.onnx, and Path::strip_prefix keeps the leading . component, so no entry matched the pinned directory (local_assets.rs). Existing installs are unaffected because they skip extraction.

Extraction now strips a leading current-directory component before matching. The size and checksum checks on each pinned file are unchanged.

Reviewer-reproducible example

On macOS, with the published archive (sha256 17f945007b52ccd8b7200ffc7c5652e9e8e961dfdf479cefcabd06cf5703630b):

cd src-tauri
cargo build -p berd-call --bin berd-call
./target/debug/berd-call models parakeet install --store-root "$(mktemp -d)/store"

Before this change the last line is berd-call models.parakeet.install failed: local voice asset archive omitted a pinned file. After it, the result is "outcome":"installed","ready":true with model.int8.onnx, tokens.txt and MODEL_LICENSE.txt under native-voice-v2/stt. I ran both on a Blox macOS workstation.

Validation

cargo test -p berd-call --lib local_assets::tests::archive_extraction_accepts_entries_under_a_leading_current_directory
  before the fix: FAILED (local voice asset archive omitted a pinned file)
  after the fix:  ok
cargo test -p berd-call --lib local_assets     13 passed
cargo fmt -p berd-call -- --check              pass

The full berd-call suite has two failures on Blox (outbound::tests::acknowledged_suspension_does_not_exhaust_the_drain_deadline, siri::tests::exact_uninstalled_voice_is_rejected_without_synthesis) that fail identically without this change.

🤖 Generated with Claude Code

[morgmart]

🤖 Automated code review

APPROVE. Fresh static Engineering review of the exact prepared three-dot comparison found no publishable changed-line defects. The extraction change removes only one leading current-directory component before applying the existing pinned archive-directory and exact-file checks, and the discriminating regression test covers the changed behavior. Supplied GitHub evidence is structurally valid and reports successful check runs at the reviewed head SHA; required checks still govern merge readiness.

Deterministic publication result: 0 blocking and 0 non-blocking inline finding(s) publishable; 0 duplicate(s) suppressed; 0 blocking screenshot-evidence requirement(s) in this review body.

Pending checks: 1 check(s) are not complete.

This approval reflects the completed code review only; merge readiness remains governed by the repository's required checks.