Repository navigation
Ziki 0.6.2 — Built-in Microphone with Bluetooth Headsets
English
- When the system default input is a classic Bluetooth headset, Ziki now records from the Mac's built-in microphone, selected by device UID, and the headset stays an output-only device.
- The notice that a Bluetooth microphone lowers playback quality is removed along with the code behind it. Playback quality no longer drops while you dictate.
- Ziki never changes the system default input, so other apps keep their own microphone selection. On a Mac without a built-in microphone, the system default input is used as before.
- Mute system output while recording is unchanged and still applies to the current output device.
- New tests cover the input routing policy for classic Bluetooth and non-Bluetooth default inputs.
- If the selected microphone stops returning audio, or never starts, dictation now reports a failure instead of quietly recording silence.
The mute Ziki writes before capturing shares hardware with a Bluetooth headset's own microphone. That is why a headset recording could previously contain no voice at all, which this release fixes at the routing level instead of explaining it to the user.
This remains an Apple Silicon, self-signed preview and is not Apple-notarized.
简体中文
- 系统默认输入是经典蓝牙耳机时,Ziki 改用 Mac 内建麦克风录音(按设备 UID 选择),耳机只作为输出设备。
- 删除“蓝牙麦克风会降低播放音质”的提示及其实现,听写期间播放音质不再下降。
- Ziki 不改变系统默认输入,其他应用仍按自己的选择使用麦克风;没有内建麦克风的机型沿用系统默认输入。
- “录音时静音系统声音”功能不变,仍然作用于当前输出设备。
- 新增输入路由策略测试,覆盖经典蓝牙与非蓝牙默认输入。
- 选定的麦克风如果没有开始或中途不再返回音频,听写会直接报错,而不是安静地录到空音频。
Ziki 录音前写入的耳机静音与耳机自身麦克风共用同一套硬件,这正是此前蓝牙录音里完全没有声音的原因;本次从采集路由上修好它,而不是把这件事故意解释给用户听。
此版本仍为 Apple Silicon 自签名预览版,尚未经 Apple 公证。
Verification
swift run ZikiCoreTestHarness: 91/91.swift run ZikiAppTestHarness: 14/14.swift scripts/verify-brand.swift: passes.- Hardware check on a Sony WH-1000XM4 as the system default input: with the headset muted, the built-in microphone still captured audio. A recording that previously used the headset microphone contained 18.8 seconds of all-zero PCM.
- The capture class used by dictation was driven on the new device route against real hardware: 2.5 seconds produced 40,107 frames of 16 kHz PCM with a non-zero signal level.
- Mute system output while recording is unchanged: live mute and restore still pass and the output volume value is untouched.