fix: prefer premium/enhanced voices over compact in Apple TTS#1117
Conversation
Previously, voiceIdentifier(for:) always returned the compact (lowest quality) voice first, ignoring any Enhanced or Premium voices the user may have downloaded via System Settings. Now collects all matching voices for the locale and picks the highest quality one: premium (2) > enhanced (1) > others including compact (0). The two-pass loop is also simplified into a single pass. Fixes tisfeng#1116
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
|
Thanks your PR, we will review it later |
There was a problem hiding this comment.
Pull request overview
Updates Apple TTS voice selection to prefer higher-quality system voices for a given language/locale.
Changes:
- Collects all voices matching the target locale and assigns a quality-based priority.
- Selects the highest-priority voice (premium > enhanced > others) instead of preferring compact voices.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
|
This change isn’t ideal. I tested an English sentence, and it uses “com.apple.speech.synthesis.voice.Albert” as the bestVoice. Albert is a quirky‑sounding voice, which isn’t suitable for general‑purpose TTS. 
|
Update Apple speech voice selection to rank premium, enhanced, then compact voices so playback falls back to compact instead of Eloquence or novelty voices. Add unit tests for the voice priority rules and register the new test file in the project.
Reflow the content-type mismatch conditions and the response data binding in stream result updates without changing behavior.
|
I have fixed this issue in 1ba6f8a Currently, the voice selection strategy is: premium > enhanced > compact, using compact as a fallback. |
|
Please review, if no problem, i will merge it |
|
Thanks for the update. I reviewed the latest changes and they look good to me. The |
Summary
Fixes #1116
voiceIdentifier(for:)previously returned the compact (lowest quality) voice immediately, ignoring any Enhanced or Premium voices the user may have downloaded via System Settings.Changes
premium(2) >enhanced(1) > others includingcompact(0)Before / After
Before: always returns compact voice first, ignores downloaded Enhanced/Premium voices.
After: prefers
premium→enhanced→compact/others. The hardcoded fallback"com.apple.voice.compact.en-US.Samantha"is intentionally kept — compact is always pre-installed and serves as a safe last resort when no voice is found for the locale.Test plan