Three tools, three different jobs
| Tool | What it changes | Main controls | What it will not do |
|---|---|---|---|
| Compressor | Reduces level only above the threshold, by the ratio, so loud and quiet words sit closer together [1][5] | Threshold, ratio, attack, release, knee, make-up gain | Set a final loudness or guarantee a peak ceiling |
| Limiter | Stops peaks from exceeding a set maximum [2][6] | Threshold or ceiling, release, lookahead | Even out a whole performance without sounding squashed |
| Peak normalization | One gain change so the loudest sample hits a level, -1 dB by default in Audacity [3] | Target peak, DC offset removal | Make two clips sound equally loud |
| Loudness normalization | One gain change to reach a perceived loudness in LUFS [4] | Target LUFS or RMS, mono handling | Stop the new peaks from clipping |
How a compressor works, with numbers
The threshold is the level above which compression applies; below it the ratio is effectively 1:1. [5] The ratio says how much the overshoot is reduced. With a -18 dB threshold and a 3:1 ratio, a word peaking at -6 dB is 12 dB over, comes out 4 dB over at -14 dB, and has been turned down 8 dB. At OBS's default 10:1 the same word would come out 1.2 dB over, at -16.8 dB, which is much harder control. OBS describes 2:1 as weak compression and 6:1 as much stronger. [5]
Attack is how quickly gain reduction kicks in after the signal crosses the threshold, and release is how quickly it lets go after the signal drops back. [1][5] Audacity notes that because attack applies compression gradually, the output briefly overshoots, and because release lets go gradually, the output briefly undershoots; and lookahead starts compressing before the threshold is reached, with latency compensated. Knee width sets how gradually compression engages around the threshold: at 0 dB it engages fully at the threshold. [1]
Because compression lowers the loud parts, the track is quieter overall afterwards. Make-up gain, 0 dB by default in Audacity, adds level back. [1] In this chain, loudness normalization does that job instead, so you can leave make-up gain at 0.
The limiter is the ceiling
A limiter is a compressor with a very high ratio and a hard maximum. Audacity describes it as passing signals below the set level and preventing stronger peaks from exceeding it, usually as the last mastering stage. Its presets include a Master Limiter at -1 dB and a VO Limiter for dialogue. Audacity warns that a long release time tends to lose quiet sounds that come after loud ones. [2]
OBS says the limiter should be the last filter in the chain and ships with a -6 dB threshold and a 60 ms release by default. [6] For a file you will publish, set the ceiling to your platform's peak limit rather than leaving that default.
Peak normalization versus loudness normalization
Audacity's Normalize sets the peak amplitude, defaults to -1 dB, removes DC offset, and will not accept positive values. Its manual advises using Normalize only to remove DC offset right after recording or importing, and leaving level changes until just before export. [3] Peak normalization says nothing about how loud a voice sounds, which is why two clips at the same peak can feel very different.
Loudness Normalization measures perceived loudness instead. Its default target is -23 LUFS, the EBU broadcast figure, which Audacity notes produces audio at about 25% of full scale, quiet for online video. It also offers an RMS mode defaulting to -20 dB, and a Treat mono as dual-mono option, on by default, so mono voice tracks measure like stereo ones. [4] For podcasts, Apple asks for about -16 LKFS integrated, plus or minus 1 dB, with true peaks no higher than -1 dBFS, measured to ITU-R BS.1770-5. [7] Change Audacity's -23 default before you run it.
The order to apply them
- 1. Clean up first: remove DC offset (Normalize with no level change), noise and clicks. [3]
- 2. Compress the voice so loud and quiet passages sit closer together. OBS places the compressor at or near the start of a filter chain. [5]
- 3. Loudness-normalize to the target: -16 LUFS for a podcast following Apple's spec. [7]
- 4. Limit last, with the ceiling at -1 dB, to catch peaks the gain change pushed up. [2][6][7]
- 5. Measure the finished file again. Any stage after the measurement can change the result, so re-run loudness analysis and nudge if you have drifted outside plus or minus 1 dB.
Starting settings for a single voice
These are starting points drawn from vendor defaults and descriptions, to adjust by ear on your voice and microphone. They are not rules.
| Setting | Start at | Basis |
|---|---|---|
| Compressor threshold | -18 dB, then raise or lower until loud words get a few dB of reduction | OBS default [5] |
| Compressor ratio | Between 2:1 and 6:1 for natural speech; 10:1 is heavy | OBS description and default [5] |
| Attack and release | 6 ms and 60 ms | OBS defaults [5] |
| Make-up gain | 0 dB, let loudness normalization set level | Audacity default [1] |
| Loudness target | -16 LUFS integrated for podcasts | Apple Podcasts [7] |
| Limiter ceiling | -1 dB | Audacity Master Limiter preset; Apple true-peak limit [2][7] |
| Mono voice tracks | Leave Treat mono as dual-mono checked | Audacity default [4] |
Signs you have overdone it
Apple warns that heavily compressed and amplified audio may be too loud, lack dynamic range and introduce distortion, and that very low-level audio makes speech hard to hear. [7] Audacity notes that compression changes the waveshape, that perceivable distortion is generally to be avoided, and that lookahead or release settings can create a pumping effect. [1]
- Breaths and room noise get as loud as words between sentences: lower the ratio or raise the threshold.
- The voice sounds flat and tiring after a minute: back off the ratio.
- The limiter is constantly working: reduce the gain going into it, or compress a little more earlier.
- Volume swells up after each pause: shorten the release.
Questions before you decide
What is the difference between a compressor and a limiter?
A compressor reduces the level of audio above a threshold by a ratio, which narrows the gap between loud and quiet words. A limiter enforces a maximum level that peaks cannot exceed, and OBS and Audacity both place it last in the chain.
Should I normalize before or after compression?
After. Compress first so the voice is consistent, then loudness-normalize to your target, then limit last. Audacity advises making level adjustments as a final step before export.
What loudness should a spoken-word podcast be?
Apple Podcasts asks for about -16 LKFS integrated, plus or minus 1 dB, with true peaks no higher than -1 dBFS. Audacity's Loudness Normalization defaults to -23 LUFS, so change the target before using it.
What ratio should I use to compress a voice?
OBS describes 2:1 as weak compression and 6:1 as much stronger, with 10:1 as its default. A ratio between 2:1 and 6:1 is a reasonable starting range for natural speech, adjusted by ear.
Sources & verification
Product details and prices can change. Check the linked provider before buying.
- Audacity Manual: Compressor Accessed 30 Oct 2026
- Audacity Manual: Limiter Accessed 30 Oct 2026
- Audacity Manual: Normalize Accessed 30 Oct 2026
- Audacity Manual: Loudness Normalization Accessed 30 Oct 2026
- OBS Knowledge Base: Compressor filter Accessed 30 Oct 2026
- OBS Knowledge Base: Limiter filter Accessed 30 Oct 2026
- Apple Podcasts for Creators: Audio requirements Accessed 30 Oct 2026
Sources link directly to providers. Product buttons may use separately labeled affiliate links. Read our disclosure.