What Is Threshold in Audio Compression?
In audio, level is commonly measured in dB. A recorded signal moves up and down as someone speaks, sings, plays, or makes noise. The threshold is the line you draw across that movement. When peaks rise above the line, the compressor reacts. When the signal stays below the line, the compressor either does nothing or eases off depending on the release setting.
This is why threshold is often the first control to set. The same ratio can feel gentle or extreme depending on where the threshold sits. A low threshold catches more of the performance; a high threshold catches only the loudest peaks. The right setting depends on the input level, not just the type of source.
How Threshold Works with Ratio
Threshold decides when compression starts. Ratio decides what happens after it starts. For example, imagine threshold is -18 dB and ratio is 4:1. If a word peaks above -18 dB, only the part above -18 dB is compressed. The audio below that point is not reduced in the same way.
If the signal reaches -10 dB, it is 8 dB over the threshold. At 4:1, that 8 dB excess becomes about 2 dB above the threshold after compression. The output peak would land around -16 dB before makeup gain. For more ratio math, see Audio Compressor Ratio Explained.
| Example setting | Signal behavior | Compressor result |
|---|
| Threshold -18 dB, ratio 4:1 | Voice peaks 8 dB over threshold | Excess is reduced to about 2 dB |
| Same threshold, ratio 2:1 | Voice peaks 8 dB over threshold | Excess is reduced to about 4 dB |
| Threshold raised to -10 dB | Most words stay below threshold | Little or no compression occurs |
High Threshold vs Low Threshold
Threshold placement changes how much of the performance is touched. A high threshold can be useful when you only want to tame occasional peaks. A lower threshold can help uneven speech, but it can also make the whole recording feel pressed down if it is set too aggressively.
| Threshold position | What happens | Typical result |
|---|
| Too high | Only rare peaks cross the line, or none do | The compressor seems inactive |
| Well placed | Louder words or peaks trigger gain reduction | Peaks are controlled while natural movement remains |
| Too low | Most of the signal stays above threshold | Constant compression, dullness, pumping, or unnatural density |
How to Set Threshold for Voice
Voice is a good place to learn threshold because the goal is usually practical: control louder words without removing the natural emphasis of the speaker. Do not begin by copying a fixed threshold from someone else, because their microphone gain and recording level may be completely different from yours.
- Play the loudest natural section of the recording, not the quiet intro.
- Lower the threshold until the gain reduction meter starts moving on louder words.
- Use about 3-6 dB of gain reduction on peaks as an initial range for speech.
- Listen at matched loudness so makeup gain does not trick you into thinking louder is better.
- Raise the threshold if every syllable sounds clamped or the room noise comes forward.
Threshold for Different Sources
Different sources need different threshold behavior. Podcast voices often need the compressor to respond to emphasis and laughter. Voice-over may need less visible compression because the delivery is usually controlled. A music bus can become lifeless if the threshold is low enough to compress every beat.
Noisy recordings need special care. Lowering the threshold can make the voice more even, but makeup gain may also raise room tone, hum, or fan noise between phrases. In that case, clean obvious noise first and use lighter compression.
| Source | Threshold approach | Watch out for |
|---|
| Podcast voice | Set for gain reduction on louder phrases | Pumping between words or laughter that ducks too hard |
| Voice-over | Higher threshold with gentle reduction | Narration losing expression |
| Streaming mic | Threshold low enough for sudden peaks | Keyboard noise or room tone becoming more obvious |
| Music bus | Higher threshold or very gentle movement | Whole mix flattening when compression never releases |
| Noisy recording | Use less compression after cleanup | Noise floor rising with makeup gain |
Threshold Does Not Mean Target File Size
Threshold is not a target size field. Setting threshold to -18 dB does not mean the file will become 18MB, and lowering threshold will not solve an upload limit. It only changes where dynamic compression begins.
If you need a smaller output, choose a file-size workflow such as Compress Audio to 10MB or Compress Audio to 16MB. Those pages estimate bitrate from the target MB and duration, which is the correct mechanism for file size.
Threshold controls when dynamic compression starts. If your upload problem is file size, compress the audio file by target MB instead.