#Recording Quality & Settings

Fine-tune Clearminutes' audio capture to match your environment and hardware. Settings affect transcription accuracy, processing speed, and file size.

#Sample Rate

Clearminutes records at 48 kHz (48,000 samples per second) — the professional standard for audio and Whisper transcription.

#Why 48 kHz?

• Professional standard: Matches industry recording, broadcasting, and post-production workflows
• Whisper compatible: Optimized for Whisper's training data and accuracy
• Headroom: Captures frequencies up to 24 kHz (human hearing tops out at ~20 kHz, so plenty of margin)
• File size: 48 kHz = ~11.5 MB per minute of mono audio

#Can I Change the Sample Rate?

No — Clearminutes locks sample rate at 48 kHz. This ensures consistent, high-quality transcription across all devices.

#Microphone Gain

Microphone gain controls the input level of your microphone before transcription. Too quiet = words missed. Too loud = audio distortion.

#How to Adjust

1. Go to Settings (⌘, on macOS, or gear icon)
2. Navigate to Audio section
3. Find Microphone Gain slider
4. Adjust the percentage (0–200%)

[SCREENSHOT: Microphone Gain slider in Settings]

#Optimal Level

The ideal level has:

• Peaks (loud parts of speech) at -6 dB to -12 dB (on a dB meter, if visible)
• Quiet speech at -20 dB to -18 dB
• No clipping (red zones in the meter)

In practical terms, when speaking normally at your mic's distance:

• Green meter at 50–75% of full scale = good level
• Red meter (clipping) = too loud, lower gain
• Grey/silent meter = too quiet, raise gain

#When to Adjust Gain

#Gain Levels by Microphone Type

#Testing Your Gain

1. Set gain to your target (e.g., 100%)
2. Start a test recording (Settings → Test Recording, or just use your next meeting)
3. Speak at your normal distance from the mic (6–12 inches)
4. Stop recording and play it back
5. Listen for:
   • Clear, natural volume (not too quiet, not distorted)
   • Minimal background noise
   • No crackling or clipping sounds
6. Adjust and repeat until satisfied

#Noise Suppression

Noise suppression removes or reduces background noise: fan hum, air conditioning, keyboard clicks, car noise, etc.

#Echo Suppression (Pro)

Echo suppression detects when the same audio appears twice in your recording — common when system audio leaks into your microphone. When enabled, it removes the duplicate so your transcript doesn't contain repeated text.

1. Go to Settings → Audio → Echo Suppression
2. Toggle ON to enable
3. Best used when recording system audio alongside your microphone

#Enable Noise Suppression

1. Go to Settings → Audio → Noise Suppression
2. Toggle ON to enable
3. Adjust the level based on your environment

#When to Use

• Quiet environment (home office): OFF or low — minimal noise to filter
• Moderate noise (AC, fan, shared office): Medium — removes hum and distant chatter
• Noisy environment (coffee shop, street): High — aggressive filtering to preserve speech clarity

If you're recording music or audio-heavy content, disable noise suppression to preserve all frequency content.

#Trade-Offs

Higher noise suppression removes more background noise but may slightly affect speech clarity. Start with medium and adjust based on your test recordings.

#Voice Activity Detection (VAD)

Voice Activity Detection is a critical feature that filters silence and non-speech audio before sending it to Whisper — reducing transcription time by ~70%.

#How VAD Works

Raw Audio Input (48 kHz)
         ↓
Microphone Gain Applied
         ↓
Noise Suppression (optional)
         ↓
VAD Filter ← ← ← Analyzes: is this speech?
         ↓ (only sends speech)
Whisper Transcription

Key insight: VAD prevents Whisper from wasting time on silence, background noise, and non-speech audio.

#VAD Sensitivity

1. Go to Settings → Transcription → VAD Sensitivity
2. Adjust the slider (0–10 scale, or Low/Medium/High):
   • 0 or Low — Very permissive, almost everything passes through
   • 5 or Medium — Balanced (default, recommended)
   • 10 or High — Very aggressive, only clear speech passes through

[SCREENSHOT: VAD Sensitivity slider in Settings]

#Understanding VAD Impact

At VAD = 0 (Low/Permissive):

• Everything gets sent to Whisper (audio pauses, background noise, paper shuffling, etc.)
• Pros: No speech is accidentally filtered
• Cons: Slower transcription (more audio to process), larger audio chunks sent to Whisper
• Use when: Speakers have very soft voices, lots of pause-filled speech (thinking out loud), need to preserve all audio nuances

At VAD = 5 (Medium/Default):

• Most speech passes through, obvious silence is filtered
• Pros: Good balance — fast transcription with no speech loss
• Cons: Very quiet speakers might be partially filtered
• Use when: Normal meetings, clear speakers, typical office environment

At VAD = 10 (High/Aggressive):

• Only loud, clear speech passes through
• Pros: Fastest transcription (70% of audio filtered), best for noisy environments
• Cons: Quiet speakers, whispers, or soft consonants might be cut off
• Use when: Noisy environment (construction, traffic, crowded cafe), need speed over completeness

#When to Adjust VAD

Lower your VAD setting if:

• You see "whispered" or "quiet speech" being missed in the transcript
• A speaker has a naturally soft voice
• Your environment is very quiet (minimal background noise)
• You're recording music, ambient sound, or audio that isn't pure speech

Raise your VAD setting if:

• Your transcript includes background noise, chatter, or paper shuffling
• You're in a noisy environment (coffee shop, open office, construction nearby)
• Transcription is too slow and you want faster results
• You want to prioritize speed over capturing every soft sound

#VAD & Processing Speed

VAD dramatically affects transcription speed. With VAD set to Medium, only ~30% of audio (the speech portions) is sent to Whisper, cutting processing time by roughly 70%. Higher VAD settings filter even more non-speech audio, further reducing processing time.

#Troubleshooting VAD

"Speech is being cut off or skipped":

• Your VAD is too aggressive
• Lower the VAD setting (move slider toward 0)
• Some speakers might have soft voices or heavy accents — adjust for them

"Background noise appears in the transcript":

• Your VAD is too permissive
• Raise the VAD setting (move slider toward 10)
• Or enable/increase noise suppression

"Transcription is very slow":

• Your VAD might be low or off
• Check VAD setting — ensure it's at 5 or higher
• Or you may need GPU acceleration (see GPU Acceleration)

#Recommended Settings by Environment

Use these as starting points, then adjust based on your first test recording.

#Quiet Home Office

Ideal for focused work, minimal distractions:

• Noise Suppression: OFF or Soft
• Microphone Gain: 80–100%
• VAD Sensitivity: Low (1–3)

Why: Quiet environments don't need suppression. Low VAD passes all speech through.

#Home Office with AC/Fan

Typical home setup with environmental noise:

• Noise Suppression: Soft
• Microphone Gain: 90–120%
• VAD Sensitivity: Medium (5)

Why: Soft suppression removes fan hum. Medium VAD filters obvious silence. 90–120% gain makes up for quiet home mics.

#Shared Office / Open Plan

Multiple people, background chatter:

• Noise Suppression: Moderate
• Microphone Gain: 70–100%
• VAD Sensitivity: Medium to High (5–7)

Why: Moderate suppression reduces chatter. Higher VAD filters more non-speech audio. Lower gain avoids picking up distant conversations.

#Conference Room (in-person meeting)

Professional meeting room, multiple participants:

• Noise Suppression: Soft (if room has AC/echo)
• Microphone Gain: 100–140% (if using built-in mic) or 70–90% (if using USB mic)
• VAD Sensitivity: Medium (5)

Why: Conference room mics vary widely. Soft suppression helps with room AC. Medium VAD balances speed and completeness.

#Coffee Shop / Outdoor

Noisy environment:

• Noise Suppression: Strong
• Microphone Gain: 100–160% (if USB mic, external speakers are loud)
• VAD Sensitivity: High (8–10)

Why: Strong suppression removes chatter and traffic. High VAD filters everything but clear speech. High gain needed because ambient noise is competing.

#Podcast / Studio Recording

Professional audio content:

• Noise Suppression: OFF
• Microphone Gain: 70–90% (use professional mic)
• VAD Sensitivity: Low (0–2)

Why: Studio audio is already noise-free. Suppression would harm quality. Low VAD preserves all audio nuances (breathing, slight pauses, etc.).

#Remote Video Call (Zoom, Teams, etc.)

Recording yourself + call participants:

• Noise Suppression: Soft to Moderate (depends on your environment)
• Microphone Gain: 80–110%
• VAD Sensitivity: Medium (5)

Why: Balanced settings for mixed audio (your voice + call audio coming through speakers). Soft-to-moderate suppression removes room noise while preserving call clarity.

#Quick Setup Checklist

Before your first important meeting:

• Set microphone gain: record 30-second test, play back, verify clear volume (not clipped)
• Enable noise suppression if your environment has AC/fan/chatter
• Set VAD to Medium (5) — adjust higher/lower after first test
• Close other apps to free RAM during transcription
• Have a USB mic handy if your environment is noisy (better than built-in)
• Test your settings on a throwaway meeting before the real one

#Related

• Configure Audio Setup →
• Choose the Right Model →
• GPU Acceleration →
• Troubleshoot Recording Issues →