Audio Compression for Streamers: Set It by Listening

A stream can be loud overall and still have unintelligible moments. Before inserting another filter, determine whether the problem is a quiet microphone, inconsistent delivery, music masking speech, or an overloaded recording. Each needs a different intervention.

What does an audio compressor do?

A downward compressor reduces louder portions according to a threshold and ratio. The knee controls the transition into compression; attack and release shape how the gain changes over time. Audacity’s compressor documentation illustrates why a static compression curve does not describe every transient.

This is dynamic-range compression. MP3 or AAC encoding reduces file size through a different process. A loudness-normalization pass measures and adjusts a program toward a chosen target; an ordinary live compressor is not proof of that target being reached.

A worked ratio example

For an idealized steady signal above a hard threshold, before makeup gain:

Output = threshold + (input − threshold) ÷ ratio

Set threshold to −18 dB and input to −6 dB. The input is 12 dB above threshold. At 4:1, that excess becomes 3 dB, so the output is −15 dB: 9 dB of gain reduction. This is arithmetic explaining the curve, not a measured OBS result or a preset recommendation. Real attack, release, knee and detector behavior change the time-varying result.

A threshold copied from another creator only makes sense relative to your incoming level. Two microphones can need different thresholds even when recording the same words.

Make a short reference recording first

Record a normal sentence, a quieter sentence, an energetic phrase and a pause. Use your usual microphone position. Save this untreated reference before adjusting filters.

Listen for distortion in the original. If the microphone or interface overloads before software receives the signal, reduce input gain and record again. A later compressor cannot restore the missing waveform. Fix obvious room noise and inconsistent distance where possible before adding makeup gain.

How should I adjust compression in OBS?

Use the microphone source’s filter controls and add Compressor, following the OBS filter reference. Begin with a modest ratio, such as 2:1, as a listening experiment. Lower the threshold gradually until louder phrases settle; keep the reference recording unchanged for comparison.

Do not replace every setting at once. If consonants lose impact, compare a slower attack. If the background audibly surges after words, revisit release and the amount of compression. These are diagnostic comparisons, not universal values for all voices.

Compare bypassed and processed versions at similar listening volume. Add only the output gain needed for the mix, then check peaks again. Automatically matching makeup gain to the largest observed gain reduction can over-boost quieter parts.

What if game audio masks my voice?

First reduce game or music level relative to speech. If you want it to dip while speaking, OBS documents sidechain ducking: the microphone controls gain reduction on another audio source. Test both speech starts and pauses so the bed does not disappear or surge distractingly.

This is separate from compressing the microphone itself. Keep a simple record of which source each filter affects. Our Twitch audio troubleshooting guide covers the wider routing and mix checks.

Where does a limiter belong?

OBS recommends placing its Limiter last in the filter chain. Recheck the complete mix after changing output gain, adding EQ boosts or changing source balance. A source-level filter does not certify every downstream stage or encoded output.

If the limiter is working hard throughout ordinary speech, revisit earlier levels instead of relying on it to repair the entire mix. Use LUFS measurement separately when evaluating a recorded program.

Can Hearably fix the stream for everyone?

The Hearably extension processes a supported viewer’s browser tab. It does not modify the stream sent by OBS or your microphone for other viewers. The online audio compressor works with uploaded files through Studio, not arbitrary live tabs.

A viewer can adjust their listening experience; a broadcaster must fix source levels and the outgoing mix. Before ending your setup session, record a short section with all sources active and listen to the actual output. Keep the untreated reference so future changes remain comparable.