Audio input: pick a mic and tune preprocessing

Input device

Settings → Audio → Input device. Lists every microphone the OS knows about. Bluetooth headsets, USB mics, built-in laptop mics, audio interfaces.

1. 1Pick a device from the dropdown.
2. 2Watch the live VU meter on the right while you speak. Bars should bounce in the green zone.
3. 3If the meter is flat, the OS gave Dimmy the wrong device (some laptops default to HDMI when a monitor is plugged in).
4. 4Click Test to record a 3-second clip and play it back. Confirms the right mic is selected.

Audio preprocessing

Dimmy preprocesses every clip before sending it to the STT engine. The defaults are tuned to match what whisper.cpp likes.

• Highpass filter at 80 Hz: removes rumble, fans, traffic. Always recommended.
• VAD (voice activity detection): skips silent stretches at the start and end of the clip. Saves cloud STT cost.
• AGC (automatic gain control): boosts quiet voices, tames loud ones. Use it if you switch between whisper-talking and normal volume.
• NaN-safe sample clamp: keeps every sample in [-1.0, 1.0]. Always on, not user-toggleable. Prevents one bad sample from crashing Whisper.

Bluetooth caveat

Bluetooth headsets often drop into the 8 kHz HFP profile when an app opens the mic, even though they support 48 kHz A2DP for output. The audio sounds tinny. Dimmy upsamples to 16 kHz before STT, which is enough for Whisper, but accuracy drops noticeably compared to a USB or built-in mic.