Recording settings on Android — input sources and audio processing
The Android app has a few settings the iOS one does not. They interact with the device and the microphone you use, so here is what each of them actually changes.
Choosing an input source
Android lets you declare what the audio is for, even from the same microphone. That declaration changes the processing the system applies on its own.
| Input source | Character |
|---|---|
| MIC | The default. A moderate amount of processing on most devices. Start here if unsure |
| VOICE_RECOGNITION | Meant for speech recognition. Little extra processing, which suits transcription |
| VOICE_COMMUNICATION | Meant for calls. Strong echo cancellation, at the cost of sound quality |
| UNPROCESSED | No processing. Whether it works depends on the device; unsupported ones fall back |
UNPROCESSED is device-dependent. A device that does not declare support ignores the request and records at roughly MIC quality. Listen to the result to find out which you got.
Leave the processing on, or switch it off
Noise suppression, echo cancellation and automatic gain all exist to make a recording easier to listen to. Whether you want them depends on the purpose.
- Transcription: off. What the processing removes can be what recognition needed
- Someone will listen back: on. Automatic gain in particular evens out the volume
- Keeping it as source material: off. Anything can be applied later
High-pass filter
This removes low frequencies. Air-conditioning hum and desk vibration sit down there, so 60 Hz or 80 Hz makes a recording noticeably cleaner. The fundamental of a human voice is generally above 100 Hz, so 80 Hz barely touches speech. At 120 Hz, a low male voice starts to sound thin.
The device decides the sample rate
You can pick 44.1k / 48k / 88.2k / 96k / 192k, but a value the device does not support silently drops to one it does. Selecting 192 kHz and actually recording at 48 kHz is entirely normal. Check the recording screen to see what is really being used.
Bluetooth microphones
With the Bluetooth preference on, audio comes from a connected headset. The call-oriented transmission is narrow-band, so it usually sounds worse than the built-in microphone, and a dropped connection stops the recording. For anything long, use the built-in microphone or a wired one.
Silence skipping and automatic splitting
Silence skipping leaves out stretches that stay below a threshold. It saves space on recordings with long idle periods, but a quiet voice can be judged as silence. For meetings and interviews, leave it off.
Turn automatic splitting on. Keeping a long session in one file means losing all of it if something goes wrong.
Recording in the background
Recording runs as a foreground service, so it continues with the screen off and while you use other apps. The notification shows elapsed time, file name and input route, which is how you confirm it is still alive. On devices with aggressive battery saving, consider excluding the app.