Problem
Exported audio files (mic and system) sound very quiet and thin.
Suspected Root Cause
Audio is captured and saved at 16 kHz / 48 kbps AAC, which limits the frequency range to 0–8 kHz (Nyquist). This narrow bandwidth makes audio perceptually quiet and lacking presence.
Specific locations:
SystemAudioCapture.swift: config.sampleRate = 16_000 → ScreenCaptureKit captures at 16 kHzMeetingRecorder.audioFileSettings: AVSampleRateKey: 16_000, AVEncoderBitRateKey: 48_000MeetingRecorder.handleSystemAudioBuffer: writes the already-downsampled 16 kHz buffer to the audio file instead of the original
Proposed Fix
- Capture system audio at 48 kHz (
config.sampleRate = 48_000)
- Write audio files at 48 kHz / 128 kbps
- In
handleSystemAudioBuffer: write the original PCM buffer to file; only use the 16 kHz converted buffer for the SFSpeech recognizer
Problem
Exported audio files (mic and system) sound very quiet and thin.
Suspected Root Cause
Audio is captured and saved at 16 kHz / 48 kbps AAC, which limits the frequency range to 0–8 kHz (Nyquist). This narrow bandwidth makes audio perceptually quiet and lacking presence.
Specific locations:
SystemAudioCapture.swift:config.sampleRate = 16_000→ ScreenCaptureKit captures at 16 kHzMeetingRecorder.audioFileSettings:AVSampleRateKey: 16_000,AVEncoderBitRateKey: 48_000MeetingRecorder.handleSystemAudioBuffer: writes the already-downsampled 16 kHz buffer to the audio file instead of the originalProposed Fix
config.sampleRate = 48_000)handleSystemAudioBuffer: write the original PCM buffer to file; only use the 16 kHz converted buffer for the SFSpeech recognizer